Welcome to this week’s deep dive into the evolving world of AI regulation and its implications. From Florida's groundbreaking law mandating AI disclaimers in political ads, to notable advancements under President Biden’s AI Executive Order, and intense debates within the EU over the handling of open-source AI models, our newsletter delivers essential insights. We also cover a significant legal challenge faced by Google, as artists assert their rights over copyrighted content used in AI training. Stay informed and ahead of the curve as we navigate these pivotal developments together. 🌐👨‍⚖️🤖

On the docket today:

• Google Sued Over AI Copyright Issues

• Florida Law Requires AI Disclaimers in Political Ads

• 180-Day Progress on Biden's AI Executive Order Achieved

• Concerns Over Open-Source AI Under the EU’s AI Act

  New here? → Subscribe

Google Sued Over AI Copyright Issues

Image created using DALL-E 3.

A coalition of visual artists has initiated a class-action lawsuit against Google, alleging that the tech giant's AI image generator, Imagen, unlawfully uses copyrighted works for its training datasets. The case, brought forth in California, challenges Google's practices around the accumulation and use of vast amounts of artistic content without permission.

Key Takeaways

• Plaintiff Details: The lawsuit was filed by a group of artists including a photographer and three illustrators, who seek to represent all copyright owners affected by Google's training practices.

• Technology in Question: Imagen, Google's text-to-image AI model, reportedly generates images by leveraging a dataset known as LAION-400M, which includes copyrighted material without the artists' consent.

• Claims Against Google: The suit claims that Imagen's training process involves direct copying of protected works, making the AI model itself an infringing derivative work.

• Google's Defense: Google contends that its AI models are trained primarily with publicly available data, and plans to challenge the artists' claims by asserting that such usage falls under fair use provisions.

• Related Litigation: This lawsuit follows similar actions taken by artists against other AI companies, underscoring a growing trend of legal challenges related to AI and copyright.

• Strategic Non-Disclosure: The complaint also highlights Google's deliberate omission of detailed dataset descriptions in its latest release, suggesting an attempt to mitigate legal risks highlighted by previous lawsuits.

• Potential Impact: The case could set significant precedents regarding copyright law and AI, influencing how tech companies approach the training of generative AI models.

Florida Law Requires AI Disclaimers in Political Ads

Image created using DALL-E 3.

Florida has enacted new legislation to address the use of artificial intelligence in political advertising. The law mandates clear disclosure when AI-generated content is used, reflecting growing concerns about transparency and the potential for misinformation.

Key Takeaways:

• Political ads using AI must include the disclaimer: “Created in whole or in part with the use of generative artificial intelligence (AI).”

• Disclaimer requirements vary by media: (1) Bold, minimum 12-point font for print, (2) Visible throughout video or TV ads, occupying at least 4% of the vertical picture height, (3) Clearly readable without user action on the internet, and (4) Audibly clear and at least 3 seconds long for audio.

• Non-compliance with disclaimer rules is a first-degree misdemeanor.

• Provisions for expedited legal complaints and hearings are included.

180-Day Progress on Biden's AI Executive Order Achieved

Image created using DALL-E 3.

Six months following President Biden's landmark Executive Order aimed at advancing and securing artificial intelligence (AI), significant strides have been made across various federal agencies. These efforts are designed to harness AI's potential while addressing its risks, ensuring equitable benefits, and maintaining American leadership in AI innovation.

Key Takeaways:

• Risk Management in AI: Agencies have addressed safety and security risks across several areas, including critical infrastructure and biological materials, and established a framework for screening synthetic nucleic acids.

• Guidance and Standards Development: Draft documents on managing generative AI risks and guidelines for critical infrastructure have been released for public comment, with final versions to offer additional guidance based on the NIST’s AI Risk Management Framework.

• AI in Workforce and Civil Rights: New principles for AI use in the workplace have been developed, alongside guidelines to ensure non-discriminatory AI applications in employment, housing, and health sectors.

• Advancing AI for Public Good: Initiatives include DOE funding for AI in science, pilots for energy management, and a sustained effort to assess AI's impact on the electric grid.

• AI Talent Development in Government: Over 150 AI professionals have been hired to support AI integration in government operations, with further plans to expand this workforce.

Concerns Over Open-Source AI Under the EU’s AI Act

Image created using DALL-E 3.

The recent developments surrounding the European Union's Artificial Intelligence Act have sparked a debate on the treatment of open-source large language models (LLMs) and their potential misuse in spreading disinformation. The legislation, nearing finalization in Brussels, grants broad exemptions to open-source AI models, which could inadvertently facilitate the creation and spread of harmful content by malicious actors. This regulatory oversight highlights a critical gap in addressing the dual nature of AI technologies as both enablers of innovation and potential tools for misuse.

Key Takeaways:

• Open-Source vs. Proprietary AI Models: Open-source LLMs allow users to modify core programming, unlike proprietary models like OpenAI's ChatGPT or Google's Gemini, which restrict user access to prevent the generation of harmful content. This openness increases the risk of misuse.

• Potential for Misuse: Tests on popular open-source models from the Hugging Face platform have shown that these models can, and often do, generate harmful content such as hate speech and misinformation with disturbing effectiveness.

• Legislative Gaps: The EU's AI Act currently exempts most open-source AI models used non-commercially from stringent regulations. This exemption fails to account for the high potential for harm, placing undue reliance on developers to follow transparency and documentation guidelines voluntarily.

• Regulatory Recommendations: Experts suggest that open-source models should be classified as high-risk AI when they pose significant threats to public health, safety, fundamental rights, or democracy. This classification would subject them to the same strict regulations as commercial AI models.

• Urgent Need for Closing Loopholes: The EU is urged to amend the AI Act to close loopholes that allow open-source AI to be exploited for disseminating disinformation. Ensuring that open-source AI models are subject to rigorous scrutiny and compliance requirements is essential to safeguard democratic values and public safety.

@lawyerswhowine

That’s all for today!

Catch ya’ll on LinkedIn

Welcome to this week’s exploration of the AI legal landscape. Whether you’re intrigued by the evolving demand in AI law careers, keen to understand the latest USPTO guidelines on AI and patent inventorship, or curious about Connecticut's ambitious plans for comprehensive AI regulation—our newsletter has you covered. 🚀👩‍⚖️

On the docket today:

• Wanna Work in AI Law? Here are Key Areas of Client Demand

• USPTO Clarifies AI's Role in Patent Inventorship

• Uber Settles Case Over AI Facial Recognition Bias

• Connecticut Contemplates Comprehensive AI Regulation

• PLUS: Canada AI Law & Policy Cheat Sheet

  New here? → Subscribe

Wanna Work in AI Law? Here are Key Areas of Client Demand

Image created using DALL-E 3.

Barry Scannell, a senior lawyer at Irish law firm William Fry, provides an overview of the evolving landscape in AI law, emphasizing the shift from theoretical considerations to tangible, practical legal advice. He outlines the current areas where clients are most actively seeking legal expertise, focusing on AI product development and regulatory compliance.

Key Areas of Legal Expertise According to Barry Scannell:

• AI Product Development: Providing guidance on the development of AI products, particularly those that incorporate large language models.

• Regulatory Compliance: Assisting with the complexities of the AI Act, focusing on compliance strategies including designations and risk classifications.

• Data Protection and Privacy: Addressing overlaps with existing regulations such as the GDPR and DSA, which are critical for the successful integration of AI technologies.

• Intellectual Property: Managing significant demands in the IP sector, particularly concerning copyright issues related to machine learning and data licensing agreements for AI applications.

USPTO Clarifies AI’s Role in Patent Inventorship

Image created using DALL-E 3.

Recent updates from the U.S. Patent and Trademark Office (USPTO) have clarified the role of AI in the patent process. While AI tools can assist in developing inventions, they have introduced new complexities for patent attorneys. These guidelines emphasize the necessity for a human to make a "significant contribution" to the invention process.

Key Takeaways:

• AI-assisted inventions are patentable if a human is identified as making a significant contribution to the invention process.

• The USPTO guidance responds to concerns by affirming that only natural persons, not software or AI systems, can be listed as inventors on patent applications.

• The Federal Circuit ruled in 2022 that AI cannot be an inventor on patents, reinforcing the need for human involvement in the inventive process.

• Attorneys are advised to carefully document the contributions of human inventors, especially when AI tools are used, to prevent potential invalidity of patents due to improper inventorship claims.

• There is an additional level of diligence required now to determine the role AI has played in the invention process, which could complicate patent applications.

• A second round of guidance from the USPTO addresses the use of AI in drafting patent applications, hinting at future complexities if AI begins to contribute autonomously to patentable ideas.

• The overarching message is that while AI can facilitate the invention process, it does not replace the essential human creative input required for patentability.

Uber Settles Case Over AI Facial Recognition Bias

Image created using DALL-E 3.

The case between Uber Eats driver Pa Edrissa Manjang and Uber regarding the use of facial recognition technology for ID checks has brought attention to potential biases in AI systems. Manjang alleged that the AI system frequently misidentified him, leading to a reduction in his work opportunities, which he claimed was a form of racial harassment.

Key Takeaways:

• AI Technology Concerns: The facial recognition system used by Uber, developed by Microsoft, has been criticized for its lower accuracy in identifying individuals of color, raising issues of racial bias in AI.

• Legal and Ethical Implications: The case underscores the need for companies to ensure AI technologies are implemented in a way that does not lead to indirect discrimination, as outlined by the U.K.'s Equality Act 2010.

• Company Accountability: Uber’s defense centered around human error and unrelated system issues, rather than flaws in the AI system itself. However, the tribunal did not dismiss the claims, indicating the complexity of distinguishing between human and AI errors in operational processes.

• Regulatory Oversight: The U.K. General Data Protection Regulation and guidelines from the Information Commissioner's Office emphasize the importance of checking biometric data processes for bias to prevent discrimination.

• Future Litigation Risks: The settlement of this case without a formal tribunal decision on AI bias leaves open questions about the fairness and reliability of AI systems in employment contexts. Future cases may further explore this issue.

• Employer Responsibilities: Companies using AI need to be transparent about their use of technology, provide avenues for human review, and ensure their systems do not disadvantage individuals based on protected characteristics.

Connecticut Contemplates Comprehensive AI Regulation

Image created using DALL-E 3.

Connecticut's legislative efforts are centering on establishing a regulatory framework for the use of AI within the state. The proposed Senate Bill 2 aims to fill the gaps left by federal inaction, addressing issues ranging from algorithmic discrimination to the creation of deep fakes, amidst Governor Ned Lamont's concerns about stifling innovation.

Key Takeaways:

• Legislative Leadership and Support: Senate Bill 2, spearheaded by Sen. James Maroney and backed by other key Senate leaders, seeks to implement standards for AI usage that could influence hiring, criminal investigations, and more.

• Governor's Skepticism: Governor Lamont has expressed reservations about Connecticut taking a prominent role in AI regulation, emphasizing the fast-moving nature of technology and the potential to hinder innovation.

• Industry Concerns: Industry representatives, including those from the Consumer Technology Association, argue that AI regulation should be handled at the federal level to avoid inconsistencies and maintain competitiveness for businesses of all sizes.

• Protection Against Discrimination: The bill proposes mechanisms to prevent algorithmic discrimination and sets a timeline for compliance by AI developers, which includes penalties for non-compliance.

• Legal Implications: The bill also explores enabling private lawsuits against AI discrimination, reflecting a strong push for individual legal recourse in the state's judiciary committee.

• Public and Political Reactions: While there is substantial support from various political figures and technology companies like Microsoft and IBM, there's also significant opposition concerned with overregulation and the practicality of state-level enforcement.

• Focus on Consumer Protection: Legislators emphasize the need to learn from past regulatory failures, such as those with internet privacy, aiming to protect consumers from potential abuses of AI technology.

• Adjustments and Amendments: Ongoing revisions to the bill show a responsive legislative process, taking into account feedback from multiple stakeholders to refine and possibly limit the scope of regulations.

PLUS: Canada AI Law & Policy Cheat Sheet

@lawyerswhowine

That’s all for today!

Catch ya’ll on LinkedIn

Hold onto your seats as we navigate through the impending surge in AI-related litigation, spotlighted by top legal minds at a University of California, Berkeley symposium. From the halls of the U.S. and European patent offices grappling with AI inventions, to the American Bar Association's insights on using AI to sharpen merger analyses, we’re here to guide you through the complexities of AI's legal landscape. And with IBM dominating 2023’s AI patent race, the stakes are higher than ever. Are you ready for the deep dive into this legal whirlwind? Let’s get started! 🚀👩‍⚖️

On the docket today:

• AI Litigation Surge Anticipated

• IBM Dominates AI Patents in 2023

• US vs. EU Patent Disclosures for AI Inventions

• Enhancing Mergers with AI

• PLUS: Senior Policy Adviser Leaks EU AI Act Revisions

  New here? → Subscribe

AI Litigation Surge Anticipated

Image created using DALL-E 3.

In a recent symposium at the University of California, Berkeley, legal experts warned of an impending wave of litigation related to AI, stemming from its expanding use and the ongoing tightening of regulations. These developments are likely to challenge courts to examine the opaque inner workings of corporate AI algorithms to assign liability.

Key takeaways include:

• Nicole A. Ozer of the ACLU highlighted that with over 30 AI-related bills under review in California, significant regulatory changes are expected by September, indicating a shift towards more civil litigation involving AI algorithms.

• A significant joint letter from multiple U.S. regulatory agencies reinforces that AI technologies are not exempt from existing laws, signaling to companies the importance of ensuring their AI systems comply with national standards to avoid potential legal consequences.

• The European Union has recently enacted the Artificial Intelligence Act, setting up the world's first regulatory framework specifically for AI algorithms, with expectations to influence global AI practices.

• Experts anticipate a rise in AI-related litigation across various domains, including intellectual property, privacy, employment decisions, and consumer disputes.

• There is a call for more precise and accurate legal language concerning AI technology, as rapid advancements in generative AI are likely to reshape legal disputes in the near future.

• Michele E. Gilman noted a lack of transparency in the use of AI by companies, leading to discriminatory decisions affecting employment, housing, and healthcare, which are often contested in small claims courts where judges may lack the expertise to adjudicate AI-related issues.

• The courts are debated on whether they should delve into the workings of AI algorithms during litigation, with opinions divided on whether understanding the technology is necessary for resolving disputes.

• Concerns are raised about the potential misuse of AI in legal settings, such as allegations of evidence being deepfakes, which could complicate litigation and drive up costs.

IBM Dominates AI Patents in 2023

Image created using DALL-E 3.

In 2023, IBM secured the top position for obtaining the most U.S. artificial intelligence patents, outpacing its nearest competitors by a substantial margin. According to a report by Harrity Patent Analytics, IBM leads with a significant lead in both AI utility patents and total AI patents held. This dominance in patent acquisition underscores IBM's robust position in the AI technological landscape and highlights the competitive efforts by other leading tech companies.

Key takeaways include:

• IBM's Leadership in AI Patents: IBM obtained 1,211 AI utility patents in 2023, making it the leader among tech companies. This is over 300 patents ahead of its closest competitor, Google, which secured 870 patents.

• Top Competitors: Following IBM and Google, Samsung, Amazon, and Microsoft also featured prominently in the list of top AI patent holders, demonstrating the breadth of involvement in AI development across different sectors of the technology industry.

• Comprehensive AI Patent Holdings: When considering the total number of AI patents held, IBM still leads with nearly 7,000 patents. Google, Microsoft, Samsung, and Amazon follow, with Intel also appearing prominently in the rankings.

• Patent Application Landscape: Samsung, IBM, and Google each have around 2,000 pending AI patent applications, with Samsung slightly leading. This indicates a strong future positioning for these companies in the AI patent space.

• Blocking Patents: Google and Microsoft are noted for holding the most patents that could block other companies' applications due to covering essential technologies or methods, with 1,145 and 1,068 blocking patents respectively.

• Impact of Blocking Patents: The companies with significant numbers of blocking patents, such as IBM, Samsung, and Microsoft, are also notably affected by others' blocking patents, which could complicate future developments and patent filings in the AI sector.

US vs. EU Patent Disclosures for AI Inventions

Image created using DALL-E 3.

Applicants seeking patent protection for AI inventions at the U.S. Patent and Trademark Office (USPTO) and the European Patent Office (EPO) face differing requirements and scrutiny. As AI technologies evolve, the focus on disclosure standards in both jurisdictions has intensified, reflecting the challenges in defining what constitutes sufficient information to secure a patent.

Key takeaways include:

• The USPTO and EPO have different criteria for the sufficiency of disclosure required for AI-related patents, influenced by their respective legal frameworks.

• In the U.S., detailed implementation specifics, such as the types of training data and the AI algorithms used, must be disclosed to demonstrate possession of the invention.

• European standards are stricter, often necessitating a detailed description of AI architecture and the technical problem solved by the invention.

• Decisions by the Patent Trial and Appeal Board (PTAB) illustrate the necessity for precise, technical disclosure to support AI patent claims, emphasizing specific algorithmic details linked to the desired outcomes.

• Several PTAB decisions have denied patent claims due to insufficient written descriptions, showing the challenges in meeting these stringent requirements.

• The EPO’s Technical Boards of Appeal (TBA) apply a more rigorous standard, often refusing patents that the USPTO might approve, due to insufficient explanations of how AI processes achieve their claimed technical effects.

• The TBA decisions underscore the importance of characterizing training data and providing clear descriptions of the AI's functional details to facilitate reproducibility by a skilled artisan.

• Both the USPTO and EPO emphasize the necessity for patent applications to explicitly detail how AI inventions achieve specific results, reflecting a broader trend towards greater specificity in patent documentation for emerging technologies.

Enhancing Mergers with AI

Image created using DALL-E 3.

At the recent American Bar Association's antitrust spring meeting in Washington, D.C., discussions focused on the integration of AI and machine learning tools in traditional merger analyses. The enhanced capabilities and methodologies that these tools provide are helping to bridge the gap between the advanced techniques used by businesses and those employed by regulatory agencies. This development marks a significant evolution in the analytical processes used in competition law enforcement across both the United States and Europe.

Key takeaways include:

• Enhanced Precision and Localization: AI and machine learning enable more granular and localized analysis in merger cases, utilizing large amounts of data from various sources such as SKU-level sales or consumer behavior tracking. This allows for more precise and geographically specific insights into the impacts of mergers.

• Improved Entry Analysis: Traditional entry analysis can be significantly enhanced using big data, offering deeper insights into how new competitors affect market dynamics and consumer behavior, which helps in understanding the competitive pressures in various local markets.

• Advanced Demand Estimation: AI tools allow the use of complex datasets to control for numerous variables, reducing the risk of omitted variable bias and allowing for a more accurate estimation of price elasticity and demand shifts due to price changes.

• Sophisticated Merger Simulation: AI can improve the realism of merger simulations by integrating dynamic and complex models of consumer response and competitive interaction, moving beyond simplistic assumptions used in traditional models.

• Detailed Analysis of Merger Efficiencies: AI techniques facilitate a deeper understanding of how mergers can optimize supply chain efficiencies and cost structures, potentially leading to direct consumer benefits such as reduced prices or improved service delivery.

• Utilization of Big Tech Techniques: The use of advanced algorithms that process vast amounts of data in real time provides a clearer picture of market dynamics and consumer preferences, which can be critical in assessing the substitutability and competition levels in various market segments.

• Need for Enhanced Data Capabilities: To effectively deploy these advanced analytical tools, agencies and their economic advisers must develop substantial capabilities in data engineering, econometrics, and understanding of both proprietary and third-party data sources.

PLUS: Senior Policy Adviser to European Parliament Leaks Revisions to AI Act

@lawyerthings_

That’s all for today!

Catch ya’ll on LinkedIn

Strap In, AI Legal Homies! Navigating the Legal Maze of AI Just Got Real! 🚀👩‍⚖️

We're slicing through the legal tangle of AI, from Google's groundbreaking French settlement to Meta's maneuver against patent trolls. As we delve into the evolving legal landscape—from protective orders in AI discovery to innovation defense strategies—hold tight. We're your guide through the twists and turns of AI's legal frontier. Ready for the whirlwind? Let’s dive in!

On the docket today:

• Google Resolves French AI Content Case

• Adapting Protective Orders for Generative AI in Legal Discovery 

• Meta Bolsters AI Innovation, Counters Trolls with Patent Transfer

• Da Vinci Partners Files Lawsuit Against AI Startup

• India Embraces AI Self-Regulation

  New here? → Subscribe

Google Resolves French AI Content Case

Image created using DALL-E 3.

Google has agreed to a settlement with the French Competition Authority, which includes a €250 million fine, over accusations of improperly using news publishers' content to train its artificial intelligence platform without adequate notification or compensation. This settlement marks a significant moment in the ongoing discourse between technology companies and content creators, particularly within the context of AI development and copyright law.

Key takeaways include:

• Fine for Non-Compliance: Google's parent company, Alphabet, and its three operating units were fined €250 million by the French Competition Authority for not adhering to a 2022 settlement related to the compensation of French publishers for using their content.

• Failure to Negotiate in Good Faith: The authority accused Google of not negotiating in good faith with publishers and not informing them or the authority about using their content to train its AI platform, Bard.

• Potential Retroactive Compensation: Following this settlement, French news publishers may pursue claims for additional, retroactive compensation for the use of their content by Google.

• Unbalanced Negotiation Conditions: According to Julien Guinot-Deléry, a Paris-based media law partner, news publishers have faced unbalanced conditions in negotiations with Google, especially regarding the valuation of indirect revenues from related rights.

• Dispute Over Proportionality of Fine: Google has contested the proportionality of the fine, stating it does not reflect the efforts made to address the concerns raised by the French Competition Authority. Despite this, Google has agreed to pay the fine and committed to making corrections to its practices.

• EU Copyright Directive and French Law: Both aim to protect news publishers' copyrights and related rights in the EU single market, requiring tech companies to negotiate payment for the use of publishers' content.

• AI Training and Copyright Concerns: The use of news content to train AI systems by Google has been highlighted as a new operation that lacked transparency, sparking debate over the application of neighboring rights to AI training.

• Le Monde's Agreement with OpenAI: Demonstrating a contrasting approach, the French newspaper Le Monde reached an agreement with OpenAI for the use of its content to train the AI platform, expected to generate significant additional revenue for the newspaper and potentially set a precedent for similar agreements in the industry.

Adapting Protective Orders for Generative AI in Legal Discovery

Image created using DALL-E 3.

The integration of generative artificial intelligence (AI) into legal processes, particularly e-discovery, introduces new risks and considerations for both the party receiving and the party producing documents. As these AI technologies, capable of analyzing and generating content based on vast data sets, become more involved in legal discovery, there is a growing need to adapt protective orders to address the unique challenges they present. This article explores how generative AI impacts the interrogation of opposing productions and suggests ways protective orders might evolve to ensure the integrity and fairness of the legal discovery process.

Key takeaways include:

• Risks for Both Parties: The use of generative AI in e-discovery poses risks not only to the party producing documents, due to potential overexposure of sensitive information, but also to the receiving party, which might inadvertently misuse AI tools, leading to ethical and legal complications.

• Need for Enhanced Protective Orders: To mitigate these risks, there is a call for protective orders to specifically address the use of generative AI. This includes clear guidelines on what AI tools can be used, how they should be used, and what measures need to be in place to protect sensitive data.

• Considerations for AI Interrogation: Parties need to consider the capabilities of AI tools in their protective orders, including their ability to infer, predict, or generate new information from the data they analyze. This requires a deeper understanding of AI technology and its implications for privacy and confidentiality.

• Balancing Innovation with Protection: While embracing the efficiency and capabilities of generative AI in legal discovery, it is crucial to balance these advantages with the need to protect sensitive information and uphold ethical standards. This balance is essential for maintaining trust in the legal process and protecting the interests of all parties involved.

Meta Bolsters AI Innovation, Counters Trolls with Patent Transfer

Image created using DALL-E 3.

Meta has initiated a strategic move to support innovation and combat patent trolls by transferring three AI-related patents to Midjourney, an image-creation research lab using generative AI. This action is intended to strengthen Midjourney's intellectual property portfolio and aid in its defense against nonparty entities (NPEs), known for their litigious pursuit of patent infringement claims. The transfer aligns with Meta's broader vision of fostering an open AI ecosystem and marks Midjourney's entrance into the LOT Network, an alliance of companies united against patent trolling.

Key takeaways:

• Strategic Patent Transfers: Meta's transfer of AI-related patents to Midjourney aims to empower the startup in its innovation efforts and provide a stronger defense against patent trolls.

• Support for Open AI Ecosystem: Jeremiah Chan, Meta's head of patents, emphasized the company's commitment to an open approach to AI, which they believe accelerates innovation and benefits society at large.

• Combating Patent Trolls: Midjourney joins the LOT Network as part of its strategy to protect against patent trolls. LOT Network's unique "license on transfer" mechanism provides its members with protection against litigation by third-party patent-asserting entities.

• Industry-Wide Protective Measures: The AI industry is proactively seeking to mitigate the risk of patent-infringement litigation by cross-licensing AI patents under the LOT license, a move designed to preemptively safeguard against disputes.

• Rise in Patent Litigation: An increase in AI patent applications has led to predictions of a surge in patent infringement lawsuits, with Norton Rose Fulbright and law firm Crowell & Moring highlighting the growing exposure to IP disputes.

• Significant Impact of Patent Trolls: Lawsuits filed by patent trolls have become a predominant form of patent litigation, posing a substantial risk to high-tech firms.

• Meta's Support for Innovation through Patent Sharing: By transferring patents to other operating companies like Starbucks and Midjourney, Meta aims to reinforce the strength of their IP rights and contribute to the innovation ecosystem.

Da Vinci Partners Files Lawsuit Against AI Startup

Image created using DALL-E 3.

Swiss patent law firm Da Vinci Partners LLC has initiated legal action against French American AI startup Kili Technology Inc. for trademark infringement and unfair competition. The lawsuit, centered on Kili Technology's use of the name "davinci" for its AI-powered patent drafting tool, raises questions about the increasingly blurred lines between goods and services in the realm of intellectual property law. Da Vinci Partners alleges that Kili Technology's naming of their product infringes on trademarks held by Da Vinci Partners and could lead to confusion within the patent attorney community, a small and tightly knit professional group.

Key takeaways include:

• Trademark Infringement Claims: Da Vinci Partners asserts that Kili Technology's use of the name "davinci" for its patent drafting AI tool infringes on its registered trademarks, DA VINCI and DA VINCI DESIGNS, held since 1996 and 2010, respectively.

• Blurring Lines Between Goods and Services: The lawsuit argues that the distinction between goods (like software) and services (like consulting) is becoming increasingly indistinct, especially with the advent of generative AI, leading to potential confusion among consumers and professionals.

• Targeted Promotion at Legal Conferences: Da Vinci Partners claims that confusion began when Kili Technology promoted the davinci tool directly to law firms at major legal conferences, such as Legalweek in New York and ABA Techshow in Chicago in early 2024.

• Impact on Patent Attorney Community: The lawsuit emphasizes the small, interconnected nature of the U.S. patent attorney community and argues that reputation among peers is of high value, making the potential for confusion particularly harmful.

• Seeking Injunction and Damages: Da Vinci Partners is seeking an injunction to prevent Kili Technology from further using the name davinci for its product, along with other unspecified damages.

Resource: Singapore AI Policy Cheat Sheet

@lawyerswhowine

That’s all for today!

Catch ya’ll on LinkedIn

Buckle Up, My AI Legal Homies! AI's Legal Odyssey Hits Warp Speed 🚀

The gavel’s coming down on tech giants, new rules are being penned across the globe, and guess what? We’re here to navigate the chaos with you. From the courtroom antics involving big tech’s data dilemmas to groundbreaking regulations from Africa to the EU, we’ve got it all. Fasten your seatbelts, it’s time to zoom through the whirlwind of AI’s legal revolution. Are you ready? Let’s goooooo!

On the docket today:

• Judge Praises AI Legal Strategy

• Ordering Pizza or Giving Up Privacy? Domino's AI Accused of Unlawful Voiceprint Collection

• Africa's AI Regulatory Leap

• Leaked: Draft Framework Convention on AI

• India Embraces AI Self-Regulation

  New here? → Subscribe

Judge Praises AI Legal Strategy

Image created using DALL-E 3.

In the Northern District of California, Judge Vince Chhabria acknowledged the plaintiffs' innovative approach in their lawsuit against Google Analytics for unauthorized data collection from healthcare provider websites. The plaintiffs included in their complaint an image showing Google Bard's response to inquiries about the appropriateness of using Google Analytics on websites covered by HIPAA and California's Confidentiality of Medical Information Act. Google Bard provided detailed explanations on why Google Analytics should not be used by healthcare providers, citing privacy concerns. Judge Chhabria found this use of AI-generated responses "very awesome," yet he also noted skepticism towards the viability of other claims in the lawsuit.

Key points:

• Google Bard's Response: The plaintiffs used Google Bard to generate responses about the use of Google Analytics on healthcare websites, highlighting potential privacy issues.

• Legal Claims: Allegations include violations under the federal Electronic Communications Privacy Act, California's Unfair Competition Law, the California Comprehensive Computer Data Access and Fraud Act, and the California Invasion of Privacy Act.

• Consent and Harm: The lawsuit raises questions about user consent and the impact of alleged privacy breaches.

Ordering Pizza or Giving Up Privacy? Domino’s AI Accused of Unlawful Voiceprint Collection 

Image created using DALL-E 3.

Domino's Pizza is facing a proposed class action lawsuit filed in the Northern District of Illinois, alleging unauthorized collection of voiceprints through its artificial intelligence telephone ordering system. The suit, which implicates both Domino's and ConverseNow Technologies Inc., raises concerns regarding the Illinois Biometric Information Privacy Act (BIPA) compliance, specifically the collection, use, and storage of biometric data without customer consent.

Key Points:

• BIPA Compliance: The lawsuit claims Domino's and ConverseNow collected voiceprints without obtaining explicit consent from customers, potentially violating BIPA requirements.

• Application of AI: The case focuses on the use of AI technology for identifying customers through voice patterns, examining the legal boundaries of technological advancements in customer service.

• Relief and Damages: Plaintiffs are seeking injunctive relief, statutory damages, and attorney fees, highlighting the legal and financial stakes of the lawsuit.

• Consumer Notification: Allegations suggest a failure to properly inform customers about the collection and use of their biometric data, as mandated by BIPA.

• Data Retention Policies: The suit criticizes the defendants' practices regarding the retention and destruction of biometric data, challenging their compliance with statutory time limits.

Africa's AI Regulatory Leap

Image created using DALL-E 3.

The Tech Hive Advisory recently unveiled a report on AI and data regulation across Africa, showcasing the continent's proactive stance in navigating the complex landscape of artificial intelligence governance. This comprehensive analysis spans 55 African nations, revealing a diverse range of regulatory approaches and commitments to data protection and AI oversight.

Highlights from the Report:

• National AI Strategies: Out of the 55 countries surveyed, five have adopted specific national strategies dedicated to AI development and governance. This indicates a strategic approach to harnessing AI for national development.

• AI Governance Bodies: Fifteen countries have taken steps to establish AI governance frameworks by forming task forces, expert bodies, agencies, councils, or committees, illustrating a commitment to structured AI oversight.

• Legislative Efforts on AI: Six nations, including Egypt, Ghana, Kenya, Nigeria, Uganda, and Zimbabwe, are at various stages of enacting AI laws, signaling a move towards comprehensive legal frameworks for AI.

• Sector-Specific Regulation: A few countries are opting for a sector-specific approach to AI regulation, with Mauritania, Nigeria, Kenya, and Tanzania leading this trend, highlighting the versatility in regulatory strategies.

Data Protection Laws

• Adoption of Data Protection Laws: A significant majority, with 37 out of the 55 countries, have enacted laws to protect data privacy, underscoring the importance of data protection in the digital age.

• Enforcement Bodies: 29 countries have gone a step further by establishing or designating bodies to enforce data protection laws, ensuring accountability and compliance.

• Protection from AI Decision-Making: Remarkably, 33 countries have included provisions in their data protection laws to safeguard individuals from being subject to algorithmic decision-making, showcasing a foresight into the potential risks of AI.

• Fairness in Data Processing: 30 nations have explicitly defined fairness as a principle of data processing within their laws, emphasizing the ethical considerations in the use of data.

Reflections:

This report highlights the dynamic and varied efforts across Africa to regulate AI and protect data. The inclusion of protections against algorithmic decision-making and the emphasis on fairness in data processing laws set a noteworthy precedent, offering lessons for Western countries in crafting their own regulations.

Leaked: Draft Framework Convention on AI

Image created using DALL-E 3.

European tech journalist, Luca Bertuzzi, has leaked a confidential draft Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law from the Council of Europe. This proposed legal instrument aims to establish common principles and rules governing activities throughout the lifecycle of AI systems in order to preserve human rights, democracy and the rule of law while promoting responsible innovation.

Key points about the draft Convention:

• It would require parties to take legislative, administrative or other measures to ensure AI activities respect human rights, democracy, and the rule of law, with the measures graduated based on assessed risks.

• It sets out principles parties must implement appropriately, such as transparency, accountability, non-discrimination, privacy protection, and respect for human dignity.

• Parties would have to identify, assess and mitigate risks AI systems pose to human rights, democracy and the rule of law, including considering potential bans or moratoria on certain AI uses.

• The Convention would directly apply to public authorities and private entities acting on their behalf.

• For other private sector AI activities, parties would need to take measures addressing risks/impacts aligned with the Convention's purposes, potentially including applying its rules to private actors.

• It provides for remedies for AI human rights violations, procedural safeguards like notification of AI system use, and oversight mechanisms.

• A Conference of Parties is established to facilitate implementation, with ability to adopt amendments and final provisions on entry into force, reservations, and dispute settlement.

• The binding treaty would be open for voluntary signature by Council of Europe members, participating non-members and the EU, requiring ratification by 5 such signatories including 3 Council of Europe states to enter into force.

India Embraces AI Self-Regulation

Imaged created using DALL-E 3.

In a significant shift from its previous regulatory stance, the Ministry of Electronics and Information Technology (MeitY) in India has issued a new advisory that encourages generative AI developers to self-regulate. This move comes as a response to evolving technologies and the need for a flexible framework that fosters innovation while addressing potential risks. Under the new guidelines, developers are no longer required to seek government approval before releasing AI models, but they are advised to label any under-tested or potentially unreliable AI-generated outputs for users in India.

Key Points:

• Emphasis on Self-Regulation: The advisory marks a departure from mandatory government approval, urging developers to adopt self-regulation practices by labeling the outputs of under-tested or unreliable AI models.

• Labeling and Consent: It is recommended that AI-generated content that could be misused, such as in creating deepfakes, should be clearly labeled to indicate its potential fallibility. A "consent popup" or similar mechanism should inform users about the unreliability of the content they are interacting with.

• Balancing Innovation and Risk: The advisory aims to strike a balance between encouraging technological innovation and mitigating the risks associated with the deployment of AI technologies. This approach reflects a more flexible regulatory environment conducive to growth and experimentation.

• Impact on AI Development and Adoption: The removal of the requirement for government approval is expected to expedite the integration and market introduction of AI models and services by major companies. This could enhance the intelligence of services and applications, promoting wider adoption of AI technologies.

• Ongoing Regulatory Evolution: Acknowledging the nascent stage of the AI ecosystem in India, the advisory suggests that regulations will continue to evolve. The government aims to adjust its approach as the technology and its applications develop further.

• Guidelines for Deepfakes and Misinformation: The advisory also addresses the concern of deepfakes and misinformation by suggesting that such content be labeled or embedded with unique metadata to identify the source of creation or modification.

• Compliance with Existing Laws: Developers are reminded that non-compliance with the IT Act 2000 and IT Rules could lead to legal consequences, underscoring the continued importance of adherence to existing legislative frameworks.

@lawyerswhowine

That’s all for today!

Catch ya’ll on LinkedIn

Holy shit ya’ll, it’s happening! The EU Parliament passed the AI act 🎉 BUT there’s more that needs to happen before it becomes law. Keep scrolling for more deets and all the other juicy updates in the legal world of AI. Like always, there’s a lot to cover. So, let’s goooooo!

On the docket today:

• EU Parliament Passes AI Act, But There’s More!

• Japan’s Text and Data Mining (TDM) Copyright Exception for AI Training

• Microsoft's AI Strategy Against Robocalls

• USPTO AI-Assisted Invention Guidance

• Judge Sanctions Lawyer $2K for AI Misuse

• Thaler vs. Perlmutter Challenges Copyright Boundaries for AI-Created Art

  New here? → Subscribe

EU Parliament Passes AI Act, But There’s More!

Image created using DALL-E 3.

Following intense lobbying from the tech industry and comprehensive negotiations among EU lawmakers, parliament passed the AI Act!

It passed with a vote of 523-46, signaling a major milestone in AI regulation. It is set for final linguistic approval and will come into effect 21 days after publication in the Official Journal, with a phased implementation timeline culminating in full enforcement by 2026.

What’s next:

• Council of Ministers' Final Vote: The bill now moves to the Council of Ministers for the final approval. Given their previous stance of support contingent on its passage in the Parliament, approval is anticipated.

• Enactment: Following approval, the bill will be signed into law, with expectations for this to occur by June.

• Implementation: The Act's provisions will be phased in gradually after it becomes law.

Japan’s Text and Data Mining (TDM) Copyright Exception for AI Training

Image created using DALL-E 3.

In a recent article, the complex landscape of Japan's TDM exception and its implications for AI training in the realm of copyright law is meticulously examined. Despite initial misinterpretations stemming from remarks made by then Minister of Education, Culture, Sports, Science and Technology, Keiko Nagaoka, Japan's Agency for Cultural Affairs (ACA) has taken proactive measures to address confusion by releasing a draft discussion paper on AI and copyright.

Key Takeaways:

• Clarification of Japan's TDM Exception: Article 30(4) of Japan's Copyright Law permits unlicensed use of copyrighted data for testing, analysis, or processing, provided it doesn't "unreasonably prejudice" copyright owners. However, it carefully distinguishes between benign data analysis and uses involving "enjoyment" of the content, where the exception doesn't apply.

• Interpretation of "Enjoyment": While the specific Japanese term is not directly translatable, it encompasses the user's benefit or perception of the copyrighted content. If the user derives benefit from the data and creates output reflecting the essential characteristics of the copyrighted work, the TDM exception does not apply.

• Impact on AI Training: Claims of Japan providing a blanket copyright exception for AI training are unfounded. The TDM exception is narrowly defined and excludes uses involving enjoyment of the content. AI developers must exercise caution, particularly regarding data obtained from sites containing pirated content, to avoid liability for infringement.

• Potential Legislative Clarity: The ACA's draft discussion paper signals Japan's commitment to clarifying its copyright laws and protecting rights-holders. While legislative changes may not be immediate, issuing official guidelines or regulations based on the paper's conclusions could provide further clarity and reaffirm Japan's adherence to international obligations.

Microsoft’s AI Strategy Against Robocalls 

Image created using DALL-E 3.

Microsoft is championing the integration of AI as a key strategy in combating robocalls and spam texts, urging the Federal Communications Commission (FCC) to recognize the potential of AI-driven solutions to enhance telecom security. By promoting AI's dual capability to both identify and mitigate fraudulent communications, Microsoft aims to shift regulatory perspectives towards the positive impacts of technological advancement.

Microsoft's dialogue with the FCC included:

• Advocating for AI in Telecom Security: Microsoft emphasized the importance of AI as a transformative tool in the telecommunications sector, particularly for detecting and mitigating robocalls and spam texts. This advocacy highlights the company's belief in AI's potential to significantly improve consumer protection against fraudulent communications.

• Presenting Azure Operator Call Protection: During its discussions, Microsoft introduced the Azure Operator Call Protection system, a new AI-based product designed to provide real-time analysis and alert users to potential scams. This product represents Microsoft's commitment to leveraging AI for enhancing the security and reliability of voice communication services without imposing additional burdens on consumers.

• Addressing Legal and Regulatory Challenges: A key aspect of Microsoft's dialogue with the FCC focused on navigating the legal challenges presented by the Wiretap Act, which currently restricts unauthorized access to the content of communications. Microsoft advocated for regulatory adjustments or clarifications that would allow AI tools to screen and block harmful communications legally.

• Emphasizing AI's Role Against Fraud: Microsoft underscored AI's dual functionality in the context of robocalls—acknowledging the risk of its misuse for creating more sophisticated scams while strongly advocating for its beneficial use in identifying and blocking such fraudulent activities. This balanced view aims to encourage the FCC to consider the positive applications of AI technology in regulatory decisions.

• Requesting Regulatory Support and Clarification: A significant part of the dialogue involved Microsoft's request for the FCC to classify AI-driven fraud detection efforts as a "necessary incident" to the provision of communication services. This classification would help clear legal hurdles related to the Wiretap Act, enabling service providers to deploy AI tools more effectively in their fight against robocalls and ensuring consumer protection.

USPTO AI-Assisted Invention Guidance

Image created using DALL-E 3.

The U.S. Patent and Trademark Office (USPTO) has recently updated its guidance on inventorship for AI-assisted inventions, clearly stating that AI entities cannot be recognized as inventors. This pivotal guidance draws a significant line between the contributions made by human inventors and the role played by artificial intelligence in the process of innovation, establishing the criteria under which individuals can be acknowledged as inventors in the age of AI-enhanced creativity.

At the core of this guidance are the Pannu factors, originating from the case Pannu v. Iolab Corp. in 1998 by the Federal Circuit, which developed a three-part test to assess significant contributions to an invention. These factors include:

• Conception or Reduction to Practice: A contributor must significantly engage in the invention's conception or its reduction to practice.

• Quality of Contribution: The contribution's quality must be substantial compared to the entire scope of the invention.

• Beyond Well-Known Concepts: A contributor's input must go beyond simply explaining well-known concepts or the current state of the art.

When applied to the context of AI-assisted inventions, the Pannu factors emphasize the necessity for significant contributions, especially underscoring the importance of an individual's input in formulating or modifying the AI's output for it to be considered as inventorship. The USPTO outlines specific criteria for inventorship in AI-enhanced innovations, including:

• Constructing Input Prompts: Crafting specific prompts that guide the AI towards a particular solution.

• Modification and Experimentation: Making significant contributions by modifying AI-generated outputs or conducting experiments to refine these outputs.

• Designing Essential Components: Developing crucial components or training the AI system tailored to solve a specific problem.

 However, certain actions do not qualify as significant contributions, such as merely presenting problems to an AI system without further intellectual input or maintaining intellectual domination over the AI's use without directly contributing to the invention's development.

The USPTO also provides illustrative scenarios to demonstrate different levels of human interaction with AI in the invention process. For example, a case involving the creation of a transaxle for a toy remote control car by two natural persons using an AI system highlights various degrees of human involvement, from no alteration of AI output, which does not qualify for inventorship, to significant modifications or the creation of a new design based on AI suggestions, which do.

These examples underscore the USPTO's stance that significant inventive contributions—like conducting experiments to alter an original design or utilizing AI to enhance a self-conceptualized design—are indispensable for an individual to be recognized as an inventor.

Judge Sanctions Lawyer $2k for AI Misuse

Imaged created using DALL-E 3.

In a recent case in Massachusetts, attorney Steven Marullo received a "restrained" sanction of $2,000 from Norfolk County Superior Court Judge Brian A. Davis for submitting court documents containing fictitious case citations generated by an artificial intelligence tool, in the context of a wrongful death case related to Sandra Birchmore. The use of the AI tool was admitted by Marullo's team, which included two recent law school graduates and an associate. Judge Davis described the submission without proper review as "categorically unacceptable," noting that ignorance of technology tools is not an acceptable excuse. Importantly, Judge Davis warned that future instances of similar misconduct would not be met with the same leniency, indicating a stricter stance on the verification of AI-generated content in legal filings.

Thaler vs. Perlmutter Challenges Copyrigt Boundaries for AI-Created Art

The case of Thaler vs. Perlmutter addresses whether AI-generated works are eligible for copyright protection, challenging existing interpretations of the Copyright Act.

• The U.S. Copyright Office argues against AI eligibility for copyright protection, emphasizing long-standing policies that restrict copyright to human authors.

• Stephen Thaler's appeal challenges the Copyright Office's decision not to register his AI-generated artwork, "A Recent Entrance to Paradise," arguing the Copyright Act does not expressly limit authorship to humans.

• The appeal highlights a broader legal and philosophical debate about the nature of creativity and authorship in the age of AI, citing examples where corporations, as non-human entities, have been granted copyright.

• The Copyright Office's brief points to the Copyright Act's language and structure, which assume an author's humanity through references to life, death, family, and legal capacities, arguing these are nonsensical when applied to machines.

• Thaler counters by emphasizing the act's purpose to incentivize the creation and dissemination of creative works for the public good, arguing that AI-generated works align with this goal by fostering innovation and expanding the range of creative works.

• This case presents a significant opportunity for the D.C. Circuit to address the intersection of copyright law and AI, potentially reshaping the legal landscape around AI-generated content.

The D.C. Circuit's upcoming decision in this case will clarify the eligibility of AI-generated works for copyright protection, significantly impacting the intersection of copyright law and artificial intelligence.

@lawyerswhowine

$800+ Worth of AI Governance Notes PLUS: IAPP’s AI Governance Course Notes

My notes for the last and final Module of the IAPP’s AI Governance is here! course is here! It’s taken a while to share all my notes. Hopefully, it was worth it and, at the very least, you have an idea of what the course covers and whether or not you want to drop a lot of cash on it. 🤓📚

Module 8: Ongoing AI Issues and Concerns – Awareness of Legal Issues

1. Developing a Coherent Tort Liability Framework for AI

• Existing Legal Frameworks: Current laws are not fully equipped to handle AI's unique challenges. This includes issues like AI's opacity, autonomous behavior, and unpredictability. There's a need to adapt traditional tort law to accommodate the complexities of AI.

• EU Initiatives: The EU Commission’s AI Liability Directive aims to compensate for AI-related damages, addressing AI's specific characteristics such as limited predictability and autonomous behavior. It also seeks to prevent companies from avoiding liability for their AI products.

• U.S. Initiatives: In the U.S., state regulations, particularly concerning autonomous vehicles, set specific safety standards. The FTC urges businesses to be transparent about AI use, ensure fair and sound decision-making, and maintain accountability.

2. AI Models and Data Licensing Issues

• Intellectual Property Rights: Defining who owns the data in AI models is a key challenge. This includes safeguarding IP rights, particularly when using third-party AI programs.

• Data Licensing Terms: These terms regulate trade secrets protection, use limitations, confidentiality, and rights assignment in AI model evolutions.

• Model Licensing: Traditional software licensing exceptions are less applicable to AI due to its evolving nature. Licensees should seek minimum performance metrics and warranties to mitigate risks of underperformance.

• Licensees’ Role: They should insist on performance metrics, warranties, thorough testing, and validation. A trial period might be beneficial to assess AI system performance.

3. Intellectual Property Rights in the Context of AI

• Examples and Challenges:

  • Copyrights: Ownership ambiguity for AI-generated works.

  • Trademarks: Issues regarding AI-created logos or slogans.

  • Patents: Recent rulings emphasize that only humans can be named as inventors.

  • Enforcement: Difficulty in detecting AI-related IP rights violations.

• Data Protection: AI's capability to scrape data poses risks for IP infringement and misuse. The legal stance on scraping for data usage is still unsettled.

Summary

The legal landscape regarding AI is rapidly evolving, with a focus on adapting existing frameworks to manage AI's unique aspects. This includes creating a coherent tort liability framework, addressing IP rights and data licensing challenges specific to AI, and implementing new directives to ensure fair and transparent use of AI technologies. These efforts are crucial for managing the risks and ethical implications of AI development and deployment, balancing innovation with legal accountability.

Module 8: Ongoing AI Issues and Concerns – Awareness of User Concerns

Educating Users About AI Systems

• AI Literacy: Focus on understanding both technological and human dimensions of AI, including impacts on privacy, agency, and proprietary information.

• Organizational Steps for AI Education:

  • Understand AI's purpose and impact on existing processes.

  • Collaborate with legal and tech teams to grasp AI limitations and legal aspects.

  • Train employees on AI use, limitations, and security/privacy controls.

  • Example: Samsung's ChatGPT misuse highlights the importance of understanding AI risks and ensuring sensitive information isn't exposed unintentionally.

Upskilling and Reskilling for AI

• AI's Impact on Employment: AI can replace or augment jobs, enhancing human capabilities in certain sectors.

• Job Automation: Roles involving repetitive tasks are more prone to automation. Creative and emotionally-driven jobs are less likely to be automated.

• Strategies for Workforce Adaptation:

  • Develop skills that complement AI, like critical thinking and creativity.

  • Encourage learning and continuous skill development.

  • Examples: Upskilling in data analysis for marketing professionals, reskilling in coding for project managers, and enhancing UX design skills for product designers.

Opt-Out Alternatives for AI

• Human Oversight in AI: Required in some laws for decision-making processes. The extent of human involvement varies based on data types and sensitivity.

• AI Program Oversight: Best practices and legal requirements often mandate human supervision of AI processes, including regular risk assessments and system monitoring.

• Bypassing AI Processes:

  • Allowing individuals to request non-AI alternatives. Availability and feasibility vary.

  • Example: U.S. Customs' alternative identity verification methods to facial recognition technology.

General Summary

• AI systems require clear user education to ensure awareness of their functions and limitations, thus preventing misuse and protecting sensitive information.

• As AI transforms the job landscape, upskilling and reskilling become essential to adapt to new demands and complement AI's capabilities.

• Providing opt-out alternatives for AI processes is increasingly important, though not always feasible. Where possible, organizations should offer clear and accessible non-AI options.

Organizations need to prioritize AI literacy, workforce adaptation strategies, and potential opt-out alternatives to address user concerns effectively. This approach ensures a responsible and informed use of AI technologies, aligning with evolving legal and ethical standards.

Module 8: Ongoing AI Issues and Concerns – Awareness of AI Auditing and Accountability Issues

Building a Global Profession of Certified Third-Party AI Auditors

• Challenges: Lack of a mature auditing framework, challenges in auditing AI due to the lack of precedents, and potential conflicts in internal audits.

• Solutions:

  • Utilize existing frameworks like COBIT, IIA AI Auditing Framework, and COSO ERM Framework.

  • Develop new auditing methods for AI incorporating data, models, outputs, and processes.

  • Promote external audits for independence, as suggested by the EU Digital Services Act.

Leveraging AI Auditing Frameworks

• COBIT 2019 Framework: Useful for defining audit scope and objectives, identifying risks, and providing assurance over AI initiatives.

• Other Frameworks: COSO ERM, GAO AI Framework, IIA AI Framework, Singapore PDPC Model AI Governance Framework, and UN Guiding Principles Reporting Framework.

• Audit Types: Includes technical, empirical, and governance audits, focusing on different aspects like data quality, bias, privacy, and security controls.

AI Audit Standards and Methodology

• Diverse Methodologies: Varied audit techniques and standards lead to different conclusions based on the audit's purpose.

• Key Areas: Includes bias audits, ethical development audits, governance, models and data evaluation, and IT general controls.

• Audit Implementation: Requires clear definitions and precise standards for effective implementation and transparency.

Enhanced Accountability for AI Systems

• Mechanisms: Development of tools and mechanisms for AI accountability and trustworthiness.

• EU AI Act: Mandates prior conformity assessments for high-risk AI systems, with a focus on risk management across the AI system’s lifecycle.

• Human Review Measures: Advisable for organizations to implement human review measures for AI decision-making.

Automating AI Governance Checks

• Benefits: Increases efficiency, consistency, accuracy, and scalability in AI deployment and compliance.

• Tools:

  • AI Verify: A framework with 11 AI ethics principles for testing AI system performance.

  • Model Card Regulatory Check: Automates regulatory compliance of AI systems, beneficial for SMEs.

• Efficiency: Automation in AI governance ensures quicker and more cost-effective compliance with evolving standards and technology.

General Summary

• The development of a global profession of certified third-party AI auditors is crucial to address the unique challenges posed by AI systems. This requires leveraging existing frameworks and developing new methodologies tailored to AI's complexities.

• AI auditing frameworks vary, focusing on different aspects like bias, ethical development, governance, and IT controls. These need to be aligned with clear standards for effective auditing.

• Enhanced accountability mechanisms are necessary, including conformity assessments and human review measures, to ensure AI systems are trustworthy and comply with legal and ethical standards.

• Automating AI governance checks is essential for maintaining productivity and adapting to evolving standards and technologies. Tools like AI Verify and Model Card Regulatory Check aid in streamlining the compliance process, making it more efficient and cost-effective.

In conclusion, the legal and ethical landscape of AI is complex and evolving. A cohesive approach to AI auditing and accountability is essential, requiring global collaboration, innovation in auditing practices, and the integration of automated tools to meet the challenges of AI governance effectively.

That’s all for today!

Catch ya’ll on LinkedIn

Hold onto your hats, AI legal nerds! ⚖️🤓 This week, we're zooming through a whirlwind of AI legal battles, from Musk's ethical showdown with OpenAI to The New York Times' copyright clash, alongside Clifford Chance's trailblazing AI integration and Dr. Vanberghen's deepfake warnings ⚠️

On the docket today:

• Musk vs. Altman

• New York Times Co. v. Microsoft Corp. et al.

• Clifford Chance's AI Integration Leap

• The AI Act and Deepfakes

• Funny Attorney Meme

• $800+ Worth of AI Governance Material

  New here? → Subscribe

Musk vs. Altman

Image created using DALL-E 3.

The complaint filed by Elon Musk against Sam Altman, Greg Brockman, OpenAI, and related entities highlights significant legal and ethical concerns with OpenAI's shift from its non-profit mission. It includes causes of action such as breach of contract, promissory estoppel, breach of fiduciary duty, unfair competition, and a demand for accounting, focusing on the secrecy surrounding GPT-4 and the development of a potentially more advanced AI model, Q*. The document also raises issues about the commercialization of AI technology and OpenAI's close ties with Microsoft, alongside governance concerns following a board coup in November 2023.

New York Times Co. v. Microsoft Corp. et al.

Image created using DALL-E 3.

OpenAI has responded to a lawsuit filed by The New York Times, accusing the company of using its content to train ChatGPT without permission. OpenAI counters by alleging that the Times engaged in hacking to produce evidence for its case, violating OpenAI's terms of use through "deceptive prompts." The AI company is seeking to dismiss most of the newspaper's claims, arguing that their use of copyrighted content falls under fair use and is essential for creating innovative products. OpenAI criticizes the Times' approach as a threat to their business model and expresses willingness to partner with news organizations, referencing an existing agreement with The Associated Press. The Times, through its lead counsel, denies OpenAI's accusations, insisting on the legitimacy of their evidence gathering and highlighting the scale of OpenAI's alleged unauthorized use of its copyrighted works. The legal battle focuses on claims of direct infringement, contributory copyright infringement, and unfair competition, with OpenAI challenging the viability and timing of these claims.

Clifford Chance's AI Integration Leap

Image created using DALL-E 3.

Clifford Chance, a British multinational law firm headquartered in London, England, is rolling out a comprehensive suite of AI-powered workplace solutions, including Microsoft’s Chat-GPT-4-powered Co-Pilot and Viva Suite, across its global workforce of nearly 7,000 individuals. This initiative follows the development of Clifford Chance Assist, a private AI tool on Microsoft’s Azure OpenAI platform, aimed at enhancing efficiency in tasks like document summarization, minute recording, and email drafting. Chief Technology Officer Paul Greenwood highlights the firm's strategic AI integration since 2019, focusing on productivity and value enhancement without foreseeing a reduction in employment levels. Clifford Chance prioritizes data control and client data protection, forbidding the use of public Chat-GPT versions for sensitive information. This move positions Clifford Chance as a leader in legal sector innovation, emphasizing the importance of AI in transforming legal practices while maintaining client trust and data security.

The AI Act and Deepfakes

Prof. Dr. Cristina Vanberghen - international legal practitioner and academic in the area of digitalisation - critically examines the escalating threat of deepfakes, emphasizing their potential to undermine democratic integrity and public trust, particularly in the context of global elections. She critiques the current regulatory measures, including the EU AI Act and the Tech Accord to Combat Deceptive Use of AI in 2024 Elections, for their lack of concrete enforcement mechanisms and their potential insufficiency in addressing the evolving nature of deepfake threats. Vanberghen advocates for the reclassification of deepfakes from “limited risk” AI systems to “high-risk” AI systems within the EU AI Act, calling for clearer legal frameworks for developer liability and the criminalization of malicious deepfake creation and dissemination. She underscores the importance of robust legal measures, international cooperation, and the empowerment of the public with digital literacy skills to effectively combat the challenges posed by deepfakes.

@lawyerswhowine

$800+ Worth of AI Governance Notes PLUS: IAPP’s AI Governance Course Notes

Module 7 is here! And, I’m just going to dump all sections here sooooo, it’s gonna be long. Get ready my fellow AI nerds! 🤓📚

Module 7: Emerging Laws and Standards in the EU

Introduction

• Context: AI governance professionals must understand the regulatory landscape in the EU regarding AI, including existing and upcoming laws.

• Focus: Three proposed AI acts in the EU are examined for their common elements and differences.

Key Topics

• AI System Classifications: Understanding the categorization of AI systems into prohibited, high-risk, limited-risk, and low-risk.

• High-Risk Systems: Grasping the specific requirements for these systems, including foundational models.

• Notification Requirements: Obligations to inform customers and national authorities about AI system aspects.

• Enforcement and Penalties: Understanding the legal consequences of noncompliance with AI regulations.

• Innovative AI Testing and Research Exemptions: Procedures and allowances for testing new AI technologies and exemptions for academic research.

• Transparency and Registration: Necessity for a public registration database for AI systems.

The EU AI Act

• Goals: To guarantee AI system safety and adherence to EU values/rights and to foster investment and innovation in AI.

• Development Timeline:

  • European Commission's initial proposal in April 2021.

  • Council of the EU's position in December 2022.

  • European Parliament's final stance in June 2023.

  • Triologue negotiations anticipated in late 2023/early 2024.

• Applicability: Extra-territorial reach, impacting AI systems marketed or used within the EU.

• Exemptions: Military, national security, defense, and R&D contexts are typically exempt from these regulations.

Framework Breakdown

• Risk-Based Approach: AI systems are categorized by risk level, each with specific regulatory requirements.

• Prohibited AI Practices: Include AI applications that:

  • Use subliminal methods to materially distort behavior causing harm.

  • Exploit vulnerabilities of specific groups (e.g., based on age or disability) to significantly influence behavior.

  • Implement social credit scoring by public authorities.

  • Utilize real-time biometric identification in public spaces for law enforcement (limited exceptions allowed).

• High-Risk AI Systems: Encompass systems with significant potential impacts on rights or safety.

• Limited Risk AI Systems: Involve AI systems like chatbots or emotional detection tools, with specific transparency obligations.

• Low/Minimal Risk AI Systems: Cover all other AI applications not specified in higher risk categories.

High-Risk AI Systems

• Areas Defined in Annex III:

  • Biometric identification and categorization of natural persons.

  • Critical infrastructure management (e.g., utilities, transport).

  • Education and vocational training (e.g., admission algorithms).

  • Employment and workforce management (e.g., recruitment, employee monitoring).

  • Essential public/private services access (e.g., credit scoring, insurance risk assessment).

  • Law enforcement applications (e.g., crime prediction, lie detection).

  • Migration, asylum, and border control management.

  • Judicial and administrative processes in justice systems.

• Requirements for Providers: Include risk management, data governance, transparency, human oversight, and robust security measures.

• Obligations for Users: Safe usage as per guidelines, incident monitoring and reporting, and ensuring human oversight.

Registration and Notification

• EU Database for High-Risk AI Systems: A publicly accessible database documenting high-risk AI systems.

• Incident Reporting: Mandates providers to report serious incidents or malfunctions to authorities within 15 days.

Key Differences Between Proposals

• Definition of AI: The Council and Parliament aimed to narrow the definition, primarily focusing on machine learning technologies.

• Expanding Prohibitions and High-Risk Categories: The European Parliament proposes to ban more AI practices and add categories to high-risk AI systems.

Summary

• Comprehensive understanding of these evolving EU regulations is crucial for AI governance professionals.

• Organizations must prepare for stringent compliance requirements, including public transparency and incident reporting.

Module 7: Emerging Laws and Standards in the EU

Introduction

• Context: Continuation of the discussion on emerging AI laws and standards in the EU.

• Focus: Understanding the evolving legal landscape for AI in the EU, including classifications, requirements, and governance frameworks.

General Purpose AI and Foundation Models

• Development: The European Commission's original proposal (April 2021) did not specifically address general-purpose AI and foundation models.

• Council's View: General-purpose AI defined as systems usable for many different purposes. Suggests future legal acts for applying high-risk AI requirements to general-purpose AI, pending consultation and impact assessment.

• European Parliament's Stance: Advocates for stringent risk assessment and mitigation for foundation models, including registration in the EU database and transparency about AI-generated content and copyrighted training data.

Governance and Enforcement Framework

• National Supervisory Authorities: Member states designate existing authorities for AI Act enforcement, possibly involving multiple organizations.

• EU Level: Creation of a European AI Board, composed of national authorities, chaired by the European Commission, ensuring uniform AI Act implementation.

• Penalties:

  • For breaching prohibited AI practices: Up to €30 million or 6% of global turnover.

  • For other non-compliances: Up to €20 million or 4% of global turnover.

  • Council proposes additional caps for SMEs to ensure proportionate fines.

The EU AI Act and Next Steps

• Identification: Determine which AI systems are classified as high-risk.

• Territorial Scope: Assess whether the AI systems fall within the AI Act's scope.

• Provider/User Clarification: Understand whether your organization is a provider or user of the AI system.

• AI Procurement Policies: Develop robust procurement processes and due diligence for AI systems.

• Gap Analysis: Compare current AI policies with AI Act requirements and address discrepancies.

• Technical Standards: Stay updated with international and European standards relevant to AI.

Summary

• Importance of Legal Teams: AI professionals should ensure their legal teams are knowledgeable about AI legislation.

• Global Impact: Like the GDPR, the EU AI Act may influence global AI regulation approaches.

• Compliance Strategy: Understanding compliance requirements is essential for AI governance teams.

Module 7: Emerging Laws and Standards Around the Globe

Introduction

• Context: Global examination of AI regulation.

• Goal: AI governance teams should leverage emerging laws globally for effective AI utilization.

• Focus: Emerging laws in various countries including Canada, the U.S., Singapore, Japan, China, and others.

Global AI Regulation Overview

• Global Activity: Countries worldwide are actively developing AI-specific legislative and regulatory frameworks.

• Approaches: Varying approaches include enhancing existing regulations and creating new AI-specific laws.

• Risk-Based Approach: Most jurisdictions are adopting a risk-based approach, except China.

• Transparency: A common element in many laws, ensuring AI systems' decisions are understandable.

Canada's AI Regulation

• AI and Data Act (C-27): Broad definition of AI, applying to private sector; focuses on high-impact systems considering harm's nature, scale of use, and consumer autonomy.

• Federal AI and Data Commissioner: Proposed for overseeing AI operations, with powers to order cessation of AI operations in certain scenarios.

Singapore's AI Framework

• AI Governance Framework: Industry-led advisory council advises on commercial AI uses.

• Model Framework (2019): Voluntary framework guiding responsible AI deployment based on transparency, fairness, and human-centric principles.

• AI Verify (2023): A toolkit supporting AI governance testing and oversight, fostering interoperability and common governance principles.

China's AI Approach

• Rights-Based Approach: Focuses on end-user rights, opt-out options, and prohibitions on price differentiation.

• Specific Directives: Target online recommendations, social media, and gaming.

• Local Governance: Municipalities passing additional AI governance features, including compliance oversight and auditing.

Other Global Perspectives

• South Africa: Protection of Personal Information Act, similar to the EU's GDPR.

• Brazil: Proposed risk-based approach with human oversight.

• India: Promotes AI use, established national committees for AI policy.

• Japan: Non-binding AI guidelines, machine learning management guidelines, cloud service guidelines for AI.

• South Korea: Non-binding guidance focusing on preventing AI dysfunction and promoting innovation.

Key Takeaways

• Review of AI Laws: AI systems are subject to existing legal standards but pose new challenges.

• Compliance Efforts: Ongoing effort to comply with developing AI laws and regulations.

• Legal Advisory: Critical for organizations to have informed legal advisors for compliance readiness.

Summary

• Proactive Regulatory Preparation: Essential for minimizing future adjustments.

• Importance of Legal Expertise: Staying informed on upcoming AI laws is crucial for compliance and operational success.

Module 7: Standards and Risk Management Frameworks - Detailed Notes

Introduction

• Focus: Designing an AI risk model suitable for intended AI use.

• Goal: Selecting aspects from various frameworks to address organization-specific AI use.

Key Frameworks and Guidelines

• ISO 31000:2018 Risk Management – Guidelines

  • Principles: Inclusive, dynamic, informed, human-centered.

  • Focus Areas: Leadership, integration, design, implementation, evaluation, improvement.

  • Process: Identify, evaluate, and mitigate risks.

• NIST AI Risk Management Framework (AI RMF)

  • Principles: Trustworthy AI characterized by validity, safety, security, accountability, explainability, privacy, and fairness.

  • Process: Identify use and risks, assess and analyze risks, prioritize and manage risks.

• Council of Europe Human Rights, Democracy, and Rule of Law Assurance Framework (HUDERIA)

  • Principles: Human dignity, freedom, prevention of harm, non-discrimination, transparency, data protection, democracy, rule of law.

  • Process: Impact assessment, risk grading, continuous monitoring.

• ISO/IEC Guide 51 Safety aspects

  • Focuses on including safety aspects in standards.

• EU Artificial Intelligence Act (EU AIA) and Singapore Model AI Governance Framework

  • Emphasizes safety, transparency, non-discrimination, environmental friendliness, transparency, explainability, repeatability, security, robustness, fairness, data governance, and accountability.

AI Governance and Responsible AI Programs

• Context: Tailoring AI governance programs to the organization's specific use cases, sector, jurisdiction, and risk tolerance.

• Frameworks vs. Principles: Frameworks operationalize values, whereas principles are a set of values.

• Key Considerations: Sector, jurisdiction, risk tolerance, business strategy, resources for implementation.

• Challenges: Managing change and cultural shifts, engaging stakeholders, raising awareness, gaining buy-in from leadership and ground teams.

• Tips: Utilize various methodologies and legal proposals to establish AI governance programs.

Components of an AI Governance Framework

• Scope: Covers AI development, procurement, and use.

• Key Elements: Risk management system, data governance, record keeping, transparency, human oversight, accuracy, robustness, cyber security, quality management, performance assessment.

• Resource Utilization: Tailoring frameworks to use cases; leveraging tools and metrics for specific AI governance issues.

Further Learning: AI Governance Framework

• ISO 31000:2018: Provides a high-level understanding of risk management principles and processes.

• NIST AI RMF: Offers practical guidance on managing risks in AI, focusing on trustworthy AI characteristics.

• HUDERIA: Suggests methodologies for impact assessments aligning with human rights principles in AI.

Challenges and Next Steps

• Change Management: Ensuring organizational alignment with AI governance principles.

• Stakeholder Engagement: Involving internal and external parties for comprehensive risk assessment and ethical AI implementation.

• Documentation: Recording decisions, risks, and mitigations for accountability.

• Resource Allocation: Prioritizing high-risk AI applications for focused evaluation and resource allocation.

• Cultural Adaptation: Embracing AI governance as an ongoing journey requiring time, effort, and consistent engagement.

Summary

• Customized Approach: No universal solution for AI risk management; tailor the approach to fit the organization's values, principles, and purposes.

• Framework Utilization: Drawing from existing frameworks and identifying organizational principles for a suitable AI risk management strategy.

• Adaptive Journey: AI governance and risk management is a continuous process, requiring adaptation and consistent evaluation.

That’s all for today!

Catch ya’ll on LinkedIn

Hey hey, AI homies! The week was filled to the brim with activity, ensuring our newsletter is packed with succulent news 💦 Let's leap directly into the action!

On the docket today:

• Sanctions for AI-Generated Fake Citations in Court Filings

• FTC Should Flag Meta's Stealth AI Training Practices

• Analyzing the Impact of Github and Tremblay Copyright Decisions

• Mistral AI, Microsoft Deal Sparks Debate

• Funny Attorney Meme

• $800+ Worth of AI Governance Material

  New here? → Subscribe

Sanctions for AI-Generated Fake Citations in Court Filings

In recent incidents, lawyers and a pro se litigant faced sanctions for citing non-existent cases in court filings, a mistake attributed to reliance on generative AI without proper verification. The Missouri case involved a pro se appellant who presented a brief with 22 of 24 cases being fictitious, resulting in a dismissal and a $10,000 fine. In Massachusetts, a lawyer filed memoranda with several fake cases, which led to a sanction of $2,000 after failing to conduct a reasonable inquiry into the accuracy of AI-generated citations. These cases highlight the legal profession's growing concern with the use of generative AI and underscore the importance of diligence in verifying AI-generated legal content.

FTC Should Flag Meta's Stealth AI Training Practices 

During Meta's Q4/2023 earnings call, CEO Mark Zuckerberg highlighted the company's vast collection of user-generated content on Facebook and Instagram, emphasizing its importance for AI training and feedback. This practice, however, raises legal and ethical concerns, particularly in light of the FTC's February 13th warning against companies that change their terms of service to use consumer data for AI training without explicit consent. Meta's privacy policy lacks clear disclosure on the use of user data for AI training, a point of contention given the FTC's stance on such practices as potentially unfair or deceptive. The absence of an opt-out option for AI training further complicates the issue, suggesting a potential violation of user privacy rights and highlighting the need for regulatory scrutiny to ensure compliance with data protection laws.

Analyzing the Impact of Github and Tremblay Copyright Decisions

The recent U.S. District Court decisions in Doe v. GITHUB, INC. and Tremblay v. OpenAI, INC. have continued to shape the legal landscape regarding copyright infringement claims against generative AI (GenAI) systems like GitHub's Copilot and OpenAI's ChatGPT.

Key Takeaways from the Github and Tremblay Decisions:

1. State Law Claims Preemption: Both decisions reaffirmed the principle that state law claims that duplicate or are equivalent to copyright claims are preempted by Section 301 of the U.S. Copyright Act. This preemption applies when the subject matter of the state law claim falls within the scope of copyright and the rights asserted under state law are equivalent to copyright rights, without adding an 'extra element' that makes the claim qualitatively different.

2. DMCA Claims on CMI: The courts have narrowed the scope of claims under the Digital Millennium Copyright Act (DMCA) related to copyright management information (CMI). In Github, the court dismissed DMCA claims by clarifying that violations require the removal or alteration of CMI from identical copies of copyrighted works, suggesting that the transformation or modification inherent in AI-generated outputs may not meet this standard.

3. Copyright Infringement and Derivative Works: The decisions continue to grapple with whether and how GenAI outputs might infringe on copyright, specifically questioning whether all outputs from GenAI systems are derivative works because they are derived from copyrighted works. The courts have yet to fully resolve this issue but have allowed cases to proceed where specific allegations of copyright output by AI were made.

4. Fair Use Defense: The unfolding decisions hint at the fair use defense's potential applicability to GenAI training and outputs, aligning with previous suggestions that the legality of GenAI systems' training under U.S. copyright law might ultimately hinge on this defense.

Mistral AI, Microsoft Deal Sparks Debate

Mistral AI's partnership with Microsoft raises significant questions about the company's involvement in shaping the AI Act and its alignment with Big Tech interests. The French AI company, previously advocating for leniency in the AI Act's foundation model rules, is now in a deal that seems at odds with its 'European champion' stance. The timing of the deal suggests a possible strategic play during the AI Act negotiations, hinting at a complex relationship with Big Tech lobbying efforts. This move also challenges France's push for 'strategic autonomy' from American tech giants, especially with Mistral's large language model being made available on Microsoft's Azure AI platform. The launch of Large, a new language model competing with OpenAI's GPT-4 but not open source, marks a shift from Mistral's open source approach, raising concerns about its position with EU policymakers as enforcement of the AI Act nears.

@lawyerswhowine

$800+ Worth of AI Governance Notes PLUS: IAPP’s AI Governance Course Notes

Module 6 is here! And, I’m just going to dump all sections here sooooo, it’s gonna be long. Get ready my fellow AI nerds! 🤓📚

Module 6: Current Laws that Apply to AI Systems

Introduction

• AI governance professionals must understand existing laws affecting AI use.

• Knowledge of these laws helps in avoiding legal/regulatory issues in AI governance and risk management.

Topics and Performance Indicators

• Unfair and Deceptive Practices Laws

• Product Safety Laws

• Intellectual Property Law

• EU Digital Services Act (transparency of recommender systems)

• Privacy Laws (data use)

Laws and AI Context

• AI falls under several regulatory frameworks.

• Compliance required throughout the AI development lifecycle.

• Categories of AI adoption: performing existing functions in new ways, and accomplishing new processes.

AI in Regulatory Context

• Existing regulatory requirements apply to AI-driven processes.

• AI introducing new processes must consider how existing laws apply.

• Highly regulated industries: financial services, health care, transportation, employment, education.

AI Evolution and Legislation

• Challenges: copyright laws application to AI, AI outputs' originality and copyright protection, human intervention in AI inventions.

• Courts, agencies, and legislators are evolving to address AI legal questions.

U.S. Legal Context

• Employment laws (Title VII, EEOC regulations)

• Consumer finance laws (Equal Credit Opportunity Act, Fair Credit Reporting Act)

• Model risk management (SR 11-7)

• Robotics safety (OSHA guidelines)

• FDA approval for software as a medical device

• FTC authority over commercial operations (privacy and security in AI)

EU Digital Services Act

• Overlaps GDPR: transparency in online platforms and advertising.

• Recommender systems and online advertising transparency.

Product Safety Laws

• EU AI Act includes product safety laws.

• U.S. Consumer Product Safety Commission developing AI standards.

• IP laws relevance in AI content generation and use.

• Data protection (GDPR, CPRA) and security standards.

Summary

• Awareness of applicable laws is essential in AI governance and risk management.

• Developing proper policies and procedures for legal compliance reduces organizational risks.

Module 6: Spotlight on the GDPR

Introduction

• GDPR addresses automated decision-making in AI.

• Compliance with GDPR is crucial for organizations using AI.

Topics and Performance Indicators

• Automated Decision-Making

• Data Protection Impact Assessments (DPIAs)

• Anonymization and its Relation to AI Systems

• AI Conformity Assessments and DPIAs

• Human Supervision of Algorithmic Systems

• Individual’s Rights in AI Logic Transparency

AI and GDPR

• Article 22: Impacts AI in automated decision-making.

• Data Protection Impact Assessments: Required for high-risk AI processing.

• Recital 26: Pseudonymization and anonymization techniques in AI.

GDPR and AI Challenges

• AI's application of GDPR principles: transparency, minimization, data subject rights.

• Implementing data subject rights in AI systems.

• AI decision-making: Requirement of manual review and challenges in black-box AI.

• Utilization of pseudonymized and anonymized data in AI.

Summary

• GDPR compliance affects AI program choices.

• Addressing GDPR considerations early avoids fines, penalties, and loss of trust.

Module 6: Liability Reform for AI

Introduction

• Awareness of changes in legal liability crucial for AI governance and risk management.

• Understanding regulators' approach to AI liability informs program design.

Topics and Performance Indicators

• EU Product Liability Law Reform

• AI Product Liability Directive

• U.S. Federal Agency Involvement (EO14091)

Product Liability Law

• Context: Increasing AI use in high-risk domains and potential serious harm.

• Challenges: Difficulty in attributing harm due to AI's autonomous nature.

• Complexity: AI's technical nature and "black box" issues complicate liability attribution.

Liability Directives

• EU proposals aim to ease proving liability in AI-caused harm.

• EU Product Liability Directive: Strict liability for AI products.

• AI Liability Directive: Fault liability linked to non-compliance with EU AI Act.

• US Legal Framework: Varies by state, with different types of liability claims.

AI Legal Framework

• US Approach: Varied state laws and minimal AI-specific case law.

• EU Approach: Directives for AI liability, enforcing compliance with AI Act.

• Federal agency guidance on AI harms (FTC, FDA, EO14091).

Summary

• Legal landscape rapidly adapting to AI.

• Organizations must be aware of AI product liabilities.

• Ensuring AI safety and compliance reduces risk of litigation and enhances reputation.

That’s all for today!

Catch ya’ll on LinkedIn

Hey hey, AI homies! This week has been packed with so much action that our newsletter is bursting with juicy updates 💦 So let's dive straight into it! 🌊

On the docket today:

• Tribunal Rejects Air Canada's Chatbot Defense

• EU AI Office Launched

• FTC Targets AI Impersonation Fraud with New Rule Proposals

• Wisconsin Implements AI Regulation in Political Campaigns and Public Sector

• FCC Mandates Consent for AI Voice Robocalls, Not Total Ban

• DOJ Advances AI Strategy

• Funny Attorney Meme

• $800+ Worth of AI Governance Material

  New here? → Subscribe

Tribunal Rejects Air Canada’s Chatbot Defense

Air Canada defended itself in a small claims case by arguing that it should not be liable for misleading information provided by its AI chatbot, claiming the chatbot was a separate legal entity responsible for its own actions. This defense came after passenger Jake Moffatt sought a refund based on inaccurate bereavement policy advice given by the chatbot. Tribunal member Christopher Rivers rejected this defense, ruling that Air Canada is responsible for all information on its website, leading to Moffatt receiving a partial refund. Following the tribunal's decision, Air Canada complied with the ruling, and the chatbot service is no longer available on the airline's website.

EU AI Office Launched!

The EU officially launched its AI Office on February 21, 2024, establishing a pivotal hub for AI expertise within the European Commission to spearhead the implementation of the AI Act, particularly focusing on general-purpose AI. 

Key highlights include:

• Central Role in Implementing the AI Act: Aims to foster the development and use of trustworthy AI, ensuring the technology's safe application across the EU.

• Trustworthy AI Development: Supports initiatives to make AI safe and reliable, upholding citizens' fundamental rights and providing legal certainty to businesses in all 27 Member States.

• Enforcement and Guidance: Will enforce rules for general-purpose AI models, develop tools for evaluating AI capabilities, and promote codes of practice in collaboration with AI developers and experts.

• International Cooperation: Aims to establish a strategic and effective European approach to AI on a global scale, promoting the EU's vision of trustworthy AI.

• Collaboration Across Sectors: Works with Member States, scientific communities, industry, civil society, and the open-source ecosystem through forums and expert groups to gather comprehensive insights on AI.

The AI Office not only marks a step forward in regulating and harnessing AI's potential within the EU but also signifies job opportunities in policy, technical, legal, and administrative roles. Interested candidates are encouraged to stay tuned for upcoming calls for expression of interest.

FTC Targets AI Impersonation Fraud with New Rule Proposals

The Federal Trade Commission (FTC) is proposing a rule to prohibit the impersonation of individuals, extending the scope of a newly finalized rule against government and business impersonation fraud. This action responds to the increasing use of AI-generated deepfakes and other forms of impersonation fraud that harm consumers. The FTC is also finalizing the Government and Business Impersonation Rule, empowering itself to pursue scammers in federal court for monetary relief. Public comments are invited on the supplemental notice of proposed rulemaking, with the final rule on government and business impersonation becoming effective 30 days after publication in the Federal Register.

Wisconsin Implements AI Regulation in Political Campaigns and Public Sector

The Wisconsin Assembly has passed two bills, AB 664 and AB 1068, to regulate artificial intelligence (AI) in political advertising and state operations. AB 664 mandates the disclosure of AI use in political ads, addressing potential issues of misinformation and election integrity. AB 1068 requires state agencies to audit their AI tools, with the goal of assessing how these technologies impact government efficiency and addressing the notion that AI could replace human employment.

FCC Mandates Consent for AI Voice Robocalls, Not Total Ban

On February 8, 2024, the Federal Communications Commission (FCC) issued a press release that, despite suggesting a complete ban on AI-generated voices in robocalls, actually clarified that such calls are subject to the consent requirements of the Telephone Consumer Protection Act (TCPA) for "artificial or prerecorded voices." This clarification aims to prevent the misuse of AI in making unsolicited robocalls to vulnerable individuals. According to the ruling, obtaining prior express consent for non-marketing calls and prior express written consent for marketing calls is mandatory. The ruling also mandates clear identification of the caller and the provision of opt-out mechanisms. This move by the FCC seeks to eliminate any uncertainties about the TCPA's applicability to AI-generated voice calls, emphasizing the protection of consumers.

DOJ Advances AI Strategy

Deputy Attorney General Lisa O. Monaco highlighted the transformative potential of AI in enhancing the Department of Justice's operations, from identifying criminals to streamlining evidence analysis. However, she also cautioned against AI's risks, such as amplifying biases, facilitating the spread of disinformation, and lowering barriers for criminal activities. The appointment of a Chief AI Officer, the launch of the Justice AI initiative, and the formation of the Disruptive Technology Strike Force are part of the DOJ's strategy to harness AI's benefits while mitigating its dangers. Monaco's remarks underscored the necessity of balancing AI's potential to improve law enforcement and public safety with the imperative to protect civil rights and maintain public trust.

$800+ Worth of AI Governance Notes PLUS: IAPP’s AI Governance Course Notes

Module 5 is here! And, I’m just going to dump all sections here sooooo, it’s gonna be long. Get ready my fellow AI nerds! 🤓📚

Module 5: Mapping, Planning and Scoping the AI Project

Introduction

• Focus: Develop AI governance and risk management program.

• Key Action: Gather stakeholders to assess risks and impacts of algorithms.

• Goal: Create methods for risk mitigation and learn documentation and communication tools.

Topics and Performance Indicators

• Define Business Case and Cost/Benefit Analysis:

  • Consider trade-offs in AI system design.

  • Assess why AI/ML is chosen over other solutions.

• Identify and Classify Risks:

  • Internal/external risks and their contributing factors.

  • Categories: Prohibitive, major, moderate.

• Construct Probability/Severity Harms Matrix and Risk Mitigation Hierarchy:

  • Systematic assessment of potential harms.

  • Develop a structured approach to risk mitigation.

• Algorithmic Impact Assessment:

  • Utilize Privacy Impact Assessments (PIAs) as a basis.

  • Tailor assessments to AI processes.

• Human Involvement in AI Decision Making:

  • Determine the level and nature of human oversight.

• Stakeholder Engagement Process:

  • Evaluate stakeholder salience.

  • Ensure diverse representation and expertise.

  • Perform positionality exercises and establish engagement methods.

• Identify AI Actors in Design, Development, and Deployment:

  • Recognize all parties involved in different AI project phases.

• Communication Plans for Regulators and Consumers:

  • Reflect compliance, disclosure obligations, transparency, and explainability.

• Chart Data Lineage and Provenance:

  • Ensure data is representative, accurate, and unbiased.

  • Use statistical sampling to identify data gaps.

• Solicit Feedback from Impacted Individuals:

  • Engage with those affected by AI systems for feedback.

• Create Preliminary Analysis Report on Risk Factors:

  • Document and manage identified risks proportionately.

Stakeholders and AI

• Define and agree on AI goals with stakeholders.

• Assess AI suitability for the mission.

• Establish success parameters.

• Determine meeting frequency for continuous evaluation.

• Establish responsibility for risks and system failures.

• Involve AI governance officers, privacy, security, procurement experts, subject matter experts, and legal teams.

Gathering Stakeholders

• Define business case and cost/benefit analysis.

• Decide on organization’s stance on AI use.

• Develop communication plan for regulators and consumers.

Identifying Risk

• Utilize tools like probability/severity harms matrix and HUDERIA risk index.

• Tailor method to organization’s needs.

• Conduct algorithmic impact assessments.

• Document risks, mitigations, and approved actions.

Summary

• Involving stakeholders, defining goals, assessing risks, and effective communication are crucial in AI implementation and ensuring ethical deployment.

Module 5: Testing and Validating the AI System During Development

Introduction

• Importance: Continuous validation to mitigate security, privacy, bias, and safety issues.

• Key Action: Identify risks to determine appropriate tests.

Topics and Performance Indicators

• Evaluate Trustworthiness and Validity:

  • Use edge cases, unseen data, and malicious input for testing.

  • Conduct repeatability assessments and complete model cards/fact sheets.

  • Create counterfactual explanations (CFEs).

• Identify Security Threats:

  • Conduct adversarial testing and threat modeling.

  • Establish multiple layers of mitigation.

  • Understand trade-offs among mitigation strategies.

• Risk Tracking and Deployment Strategies:

  • Document risk changes over time.

  • Select appropriate deployment strategies.

Testing and Validation

• Types of Testing: Accuracy, robustness, reliability, privacy, interpretability, safety, and bias assessments.

• Resource Allocation: Prioritize testing based on risks and organizational capacity.

• Compliance: Maintain documentation for audits and regulatory compliance.

• Tailoring Testing: Customize tests based on algorithm type, industry standards, and AI’s purpose.

Use Cases and Resources

• Align testing data and processes to specific use cases.

• Understand resource allocation and prioritize based on risks.

Other Testing Considerations

• Conduct adversarial testing and establish error mitigation layers.

• Document all decisions and processes.

Summary

• Thorough testing and validation ensure AI system reliability, security, and performance. Tailoring testing approaches based on risks and purpose is critical.

Module 5: Monitoring and Maintaining a System

Introduction

• Focus: Continual monitoring post-deployment.

• Key Action: Manage ongoing system performance and security.

Topics and Performance Indicators

• Security Protocols and Industry Standards:

  • Follow protocols and standards for new risks post-deployment.

  • Develop an incident response plan.

• Monitoring the System:

  • Use inventory and risk scores for AI systems.

  • Allocate resources based on risk scores.

  • Regularly review algorithms for drift and changes.

• Documenting and Responding to Incidents:

  • Maintain model versions and datasets.

  • Create challenger models for transparency.

  • Respond to risks with prioritized action plans.

Monitoring AI Systems

• Implement tests to evaluate goals achievement and unintended outputs.

• Use challenger models to predict additional risks.

AI Performance and Incident Response Plans

• Consider performance issues as incidents.

• Document model versions for accuracy and transparency.

• Address internal and external risks with a risk score system.

Summary

• Effective AI governance requires ongoing monitoring, fine-tuning, and robust incident response plans. Staying updated with best practices and documenting AI system purposes and limitations is essential.

That’s all for today!

Catch ya’ll on LinkedIn

What up AI homies?! Welcome to your dose of AI legal news. I’m keeping it light this time around because “ain’t nobody got time for…” writing a whole newsletter when you’re lawyering full time. 🙄⚖️🤖

On the docket today:

• EU Lawmakers Ratify Political Deal on Artificial Intelligence Rules

• FCC Goes One Step Further by Making AI-Generated Voices in Robocalls Illegal

• Northern District of California grants unfair competition claim again OpenAI

• EU AI Act: Prohibited Practices Acts Cheat Sheet

• Depressing Attorney Meme

• $800+ Worth of AI Governance Material

  New here? → Subscribe

Depressing Attorney Meme

$800+ Worth of AI Governance Notes PLUS: IAPP’s AI Governance Course Notes

Moving on to Module 4 of the IAPP’s AI Governance course (notes). Alrighty, it’s time to geek out and keep learning about AI Governance!!! 🤓📚

Module 4: Interoperability of AI Risk Management

Introduction

• Importance of reviewing and harmonizing existing risk management with new AI risk strategies.

• Understanding existing programs and their objectives.

• Creating new processes for unique AI risks and ensuring efficiency and protection.

Security Risk

• Risks include hallucinations, deepfakes, training data poisoning, data leakage, and filter bubbles.

• Organizations respond by banning or limiting AI use.

• AI algorithms can concentrate power, leading to overreliance and security holes.

• Risks of adversarial machine learning attacks.

• Misuse of AI and transfer learning attacks.

• Storing training data in less secure environments.

Operational Risk

• High costs in hardware and environmental impact.

• Data corruption and poisoning risks.

• Need for good identity and access management.

Privacy Risk

• Data persistence, repurposing, spillover, and collection from AI.

• Challenges in informed consent and opting out.

• Limiting data collection and deletion complexities.

Regulation and Legal Risk

• Compliance with laws and regulations.

• Liability, intellectual property, human rights, and reputational risks.

• Importance of aligning different risk management strategies to prevent gaps.

Privacy Harms

• Categories include physical, reputational, relationship, economic, discrimination, psychological, and autonomy harms.

• Use of AI in autonomous weapons, healthcare decisions, and labor displacement.

• Effects on justice system and accountability.

Summary

• Balancing AI benefits with potential harms.

• Evolving risk management strategies to include AI.

• Regular assessment of vulnerabilities and potential harms.

Module 4: Principles of AI Risk Management

Introduction

• Incorporating risk management and AI governance principles.

• Understanding stakeholders, AI programs, and associated risks.

Principles

• Pro-Innovation Mindset: Preparedness for changes and alignment with principles.

• Risk-Centric Governance: Considering risk factors in governance.

• Consensus-Driven Planning and Design: Involving all stakeholders and ensuring understanding of needs vs. risks.

• Outcome-Focused Team: Clarity on desired outcomes and exploring better achievement methods.

• Non-Prescriptive Approach: Context-specific risk approaches for adjustment and evolution.

• Law-, Industry-, and Technology-Agnostic Framework: Interoperability and flexibility without bias.

• Third-Party Risk Management: Ensuring end-to-end accountability.

Risk Management

• Treating risks case-by-case.

• Involving business and technical stakeholders.

• Enumerating potential harms and assessing data used in AI.

• Technical tools to assess AI for bias and risks.

• Categorizing AI risks according to the EU model.

Summary

• Regular risk assessments in the context of regulatory requirements, risk tolerance, and industry standards.

Module 4: Establishing AI Governance and Strategy

Introduction

• Understanding the organizational operations and incentive structures.

• Tailoring AI governance to the organization's type.

AI Governance Stakeholders

• Identifying key stakeholders and their roles.

• Engaging leadership for AI governance support.

• Addressing pressures on tech teams and influencing behavioral change.

AI Governance Structure

• Leveraging existing structures and ensuring company-wide buy-in.

• Transparency about the state of AI governance maturity.

• Defining roles and responsibilities clearly.

• Incentivizing effective and safe AI products.

• Engaging HR for role identification and success measures.

Summary

• Building AI governance starts with understanding organizational structure and culture.

• Engaging stakeholders early and fostering a culture of responsible AI.

• Iterating the governance program from conception to completion, with clear roles and incentives.

That’s all for today!

Catch ya’ll on LinkedIn

What up homies?! Welcome back to your weekly dose of AI legal news – where the future is here, and it's loaded with robots making questionable phone calls and lawmakers hustling to keep up! ⚖️🤖

On the docket today:

• FCC Clamps Down on AI-Generated Biden Impersonation Robocalls

• Implementing AI in Law Firms: Strategies Revealed

• US Introduced AI Environmental Impacts Act of 2024

• Navigating the EU AI Act: Flowchart

• PLUS: IAPP’s AI Governance Course Notes

• Hilarious Meme

  New here? → Subscribe

FCC Clamps Down on AI-Generated Biden Impersonation Robocalls

Image created using DALL-E 3

On February 6, 2024, the Federal Communications Commission (FCC) took action against Lingo Telecom, LLC, for its role in distributing robocalls using artificial intelligence to imitate President Joseph R. Biden, Jr. This move comes in response to robocalls aimed at influencing voter behavior ahead of the New Hampshire Presidential Primary by advising Democrats not to vote, under the guise of the President's voice.

Lingo Telecom, identified for originating these calls, was issued a cease-and-desist letter by the FCC. The agency highlighted the utilization of deepfake technology, an advanced AI tool capable of creating highly realistic audio and video impersonations. These robocalls misrepresented caller identification to appear as if coming from notable Democratic figures, leveraging AI to craft messages that could potentially sway electoral outcomes.

Given a 48-hour window to mitigate this illegal activity, Lingo Telecom faces potential blocking by U.S.-based voice service providers if it fails to comply. This enforcement action underscores the FCC's commitment to combating the misuse of AI in spreading misinformation and its effects on the integrity of elections.

The incident represents a significant challenge in regulating AI technologies, demonstrating their potential for harm when used unethically. As AI continues to evolve, the FCC's measures against Lingo Telecom serve as a critical step in addressing the dual-use nature of artificial intelligence, balancing its benefits against the risks of misuse and deception.

Implementing AI in Law Firms: Strategies Revealed

Image created using DALL-E 3

Implementing AI in Law Firms: Strategies Revealed

Attorneys seeking to integrate AI into their practices now have access to a distilled guide based on strategies from Winston and Strawn's AI Strategy Group, as featured in the LegalTech News. Dive into the essential steps for adopting AI effectively in your law firm, ensuring you leverage technology to its fullest potential while maintaining the high standards of legal service your clients expect.

Essential Steps:

1. Clarify Your AI Vision: Before diving into AI, define what success looks like for your firm. Consider AI's role not just in streamlining your internal processes but also in enhancing the counsel you provide to clients amidst a shifting legal landscape.

2. Plan Before You Announce: Assemble a task force and have a clear action plan ready before making any public announcements about your AI initiatives. Solid preparation will lay the foundation for smoother integration of AI into your firm's operations.

3. Create a Diverse Team: Your AI task force should bring together legal minds and tech experts. This multidisciplinary team will be better equipped to tackle the complex challenges of AI adoption.

4. Learn From Peers: Engage with other firms and AI vendors to understand their experiences and solutions. Benchmarking against peers can offer valuable insights and help you make more informed decisions about the AI tools that will best serve your firm.

5. Be Strategic with AI Deployment: Pinpoint specific areas where AI can deliver immediate benefits. Legal research and document review are prime candidates, given AI's proven efficacy in these domains.

6. Keep Costs in Mind: Be mindful of the costs associated with AI. Seek out vendors that offer solutions balancing affordability with functionality. The AI market is evolving, with more budget-friendly options expected on the horizon.

7. Choose the Right Vendor Fit: Evaluate whether a vendor's offerings align with your long-term goals. Do you need a specialized tool for e-discovery, or are you looking for a broader toolset for various document-related tasks? Decide based on your firm's specific needs.

8. Opt for User-Friendly AI: Select AI tools that are intuitive and easy for all attorneys to use, regardless of their tech savviness. User-friendly interfaces that combine robust language models with natural language processing can democratize AI use in your firm.

9. Assess AI's Capabilities: Understand AI's current capabilities and limitations. While excellent for summarizing and researching, AI may not yet adeptly handle more complex legal tasks or manage large document sets effectively. Test the tools to ensure they meet the specific demands of your practice.

US Introduces AI Environmental Impacts Act of 2024

Image created using DALL-E 3

The U.S. has taken a pivotal step toward addressing the environmental implications of artificial intelligence with the introduction of the "Artificial Intelligence Environmental Impacts Act of 2024." This groundbreaking bill, introduced by key lawmakers, seeks to blend technological innovation with environmental responsibility, paving the way for a sustainable future in the AI sector.

Key Insights from the Bill:

• Legislative Initiative: Spearheaded by Congress members Anna Eshoo, Don Beyer, and Senators Ed Markey and Martin Heinrich, the bill targets the environmental footprint of AI.

• Standards and Reporting: The National Institute of Standards and Technology (NIST) is tasked with creating standards for measuring AI's environmental impacts, including energy use, pollution, and e-waste. A voluntary framework for AI developers to report on environmental impacts is proposed, encouraging transparency.

• Comprehensive Study: A two-year interagency study, involving NIST, the Environmental Protection Agency (EPA), and the Department of Energy, will investigate AI's environmental impacts.

• Voluntary Framework: The bill introduces a voluntary approach to environmental reporting by AI developers, which could potentially lead to future mandatory regulations.

• Future Implications: Early adoption of sustainable practices by AI firms is encouraged, potentially influencing future regulatory standards. The bill underscores the importance of integrating environmental sustainability into the AI development process.

Navigating the EU AI Act: Flowchart

This flowchart is crazy…in a good way! If your organization is a provider, deployer, distributor, product manufacturer, authorized representative, or affected person, it helps you find your way, easily, through the act.

I included it as an image file below but you can access the PDF by clicking the headline.

PLUS: IAPP’s AI Governance Course Notes

Alrighty, it’s time to geek out and keep learning about AI Governance!!!

We are in Module 3: Understanding the AI Development Life Cycle. Luckily, this is a very short module and I will be able to share my notes from module 4 next week. 🚀📚

Module 3: Understanding the AI Development Life Cycle

The AI development life cycle requires meticulous planning, design, development, and implementation phases. This structured approach ensures AI projects align with organizational goals while mitigating risks and adapting to evolving needs.

Planning Phase:

• Business Objectives and Requirements: It's crucial to define the business problem at the outset. AI can address various business problems, predominantly classified into three categories:

  • Classification: Differentiating data into categories.

  • Regression: Predicting future outcomes based on past data.

  • Recommendation: Suggesting actions or choices.

• AI Use Cases: These must align with the organization’s mission. This involves identifying gaps in current processes and determining how AI can bridge these gaps.

• Project Scope and Governance:

  • Scope Determination: Prioritize use cases based on business needs, considering impact, effort, and fit with organizational goals.

  • Governance Structure: Establish who is responsible for AI governance, policy formulation, and overseeing the development and testing processes. An executive champion can facilitate broader support.

Design Phase:

• Data Strategy:

  • Data Collection: Assess the required type, amount, and source of data, whether internal or external.

  • Data Quality and Format: Focus on data quality to avoid 'garbage in, garbage out' scenarios and consider the data format (structured, unstructured, static, streaming).

  • Privacy Technologies: Implement techniques like anonymization, data minimization, differential privacy, and federated learning to enhance privacy.

• AI System Architecture and Model Selection:

  • Algorithm Choice: Select based on accuracy, interpretability, and alignment with the objectives.

  • Model Requirements: Consider constraints such as time and accuracy.

Development Phase:

• Model Building and Feature Engineering:

  • Feature Definition and Engineering: Use domain knowledge to create and select relevant features efficiently.

  • Model Training and Testing: Iteratively train, test, evaluate, and retrain models to ensure performance aligns with business objectives and can handle new data effectively.

Implementation Phase:

• Readiness Assessment and Deployment:

  • Readiness Assessment: Evaluate if the model meets quality standards and is ready for production.

  • Deployment: Plan for continuous monitoring post-deployment.

• Continuous Monitoring and Maintenance:

  • Performance Monitoring: Watch for accuracy deviations, irregular decisions, and data drift.

  • Ongoing Maintenance: Regularly update the model to address changes in data and operational environments.

In summary, the AI development life cycle emphasizes defining clear business objectives, ensuring high-quality and privacy-compliant data, careful model selection and training, and ongoing monitoring and maintenance post-deployment.

Meme Of The Week:

@airules

That’s all for today!

Catch ya’ll on LinkedIn

I’m testing out a new format this week. I’ll list some of the latest headlines in AI legal news andddd that’s it. Click on a headline if you want more deets. Let’s dive in!

On the docket today:

• International AI Treaty Debate Intensifies as Text Softens

• Attorneys Whine Over AI Regulation Proposal by U.S. Federal Court

• France Nears Approval of AI Act, Seeks Clarity on Transparency and Trade Secret

• Germany Backs EU AI Act with Compromise, Boosting Innovation and Trust in AI

• Hilarious Lawyer Meme

• PLUS: IAPP’s AI Governance Course Notes

  New here? → Subscribe

Hilarious Lawyer Meme:

@lawyerswhowine

IAPP’s AI Governance Course Notes

Alrighty, it’s time to geek out and keep learning about AI Governance!!!

We are in Module 3: Understanding the AI Development Life Cycle. Luckily, this is a very short module and I will be able to share my notes from module 4 next week. 🚀📚

Module 3: Understanding the AI Development Life Cycle

The AI development life cycle requires meticulous planning, design, development, and implementation phases. This structured approach ensures AI projects align with organizational goals while mitigating risks and adapting to evolving needs.

Planning Phase:

• Business Objectives and Requirements: It's crucial to define the business problem at the outset. AI can address various business problems, predominantly classified into three categories:

  • Classification: Differentiating data into categories.

  • Regression: Predicting future outcomes based on past data.

  • Recommendation: Suggesting actions or choices.

• AI Use Cases: These must align with the organization’s mission. This involves identifying gaps in current processes and determining how AI can bridge these gaps.

• Project Scope and Governance:

  • Scope Determination: Prioritize use cases based on business needs, considering impact, effort, and fit with organizational goals.

  • Governance Structure: Establish who is responsible for AI governance, policy formulation, and overseeing the development and testing processes. An executive champion can facilitate broader support.

Design Phase:

• Data Strategy:

  • Data Collection: Assess the required type, amount, and source of data, whether internal or external.

  • Data Quality and Format: Focus on data quality to avoid 'garbage in, garbage out' scenarios and consider the data format (structured, unstructured, static, streaming).

  • Privacy Technologies: Implement techniques like anonymization, data minimization, differential privacy, and federated learning to enhance privacy.

• AI System Architecture and Model Selection:

  • Algorithm Choice: Select based on accuracy, interpretability, and alignment with the objectives.

  • Model Requirements: Consider constraints such as time and accuracy.

Development Phase:

• Model Building and Feature Engineering:

  • Feature Definition and Engineering: Use domain knowledge to create and select relevant features efficiently.

  • Model Training and Testing: Iteratively train, test, evaluate, and retrain models to ensure performance aligns with business objectives and can handle new data effectively.

Implementation Phase:

• Readiness Assessment and Deployment:

  • Readiness Assessment: Evaluate if the model meets quality standards and is ready for production.

  • Deployment: Plan for continuous monitoring post-deployment.

• Continuous Monitoring and Maintenance:

  • Performance Monitoring: Watch for accuracy deviations, irregular decisions, and data drift.

  • Ongoing Maintenance: Regularly update the model to address changes in data and operational environments.

In summary, the AI development life cycle emphasizes defining clear business objectives, ensuring high-quality and privacy-compliant data, careful model selection and training, and ongoing monitoring and maintenance post-deployment.

That’s all for today!

Catch ya’ll on LinkedIn

Did ya miss me?! I missed you and all the craziness in AI news! Let’s dive in! This week we're covering a range of crucial topics, including the recent leak of the EU AI Act's final text by journalist Luca Bertuzzi, the impending establishment of the European Artificial Intelligence Office, and China's ambitious draft guidelines for AI industry standardization. Additionally, I include insightful notes from IAPP's AI Governance course, discussing the characteristics of trustworthy AI systems. Stay informed on these pivotal movements shaping the future of AI and law. ⚖️🤖

On the docket today:

• Final Text of EU AI Act Leaked

• The EU AI Office is Almost Ready

• China Drafts AI Standardized Guidelines

• China AI Law Cheat Sheet

• PLUS: IAPP’s AI Governance Course Notes

• Hilarious Meme

  New here? → Subscribe

Final Text of EU AI Act Leaked

Image created using DALL-E 3

On January 22, 2024, Luca Bertuzzi, a renowned journalist specializing in European affairs, made a significant move by leaking the final text of the EU Artificial Intelligence Act. His expertise in tracking the progress of this critical legislation has been invaluable in keeping the public informed.

This strategic leak occurred just before the Telecom Working Party, a key technical body of the EU Council, was set to discuss the Act, leading up to its expected formal adoption by COREPER on February 2. Bertuzzi has raised concerns that this tight schedule allows inadequate time for national delegates to conduct a comprehensive analysis of the document. This constraint necessitates a focus on the most vital articles, possibly at the expense of a holistic understanding.

Bertuzzi reports that France is actively seeking alliances with other EU countries to potentially form a blocking minority. The aim behind this diplomatic maneuvering is to delay the COREPER vote, creating an opportunity to propose and negotiate specific amendments to the text. However, as of now, France has not succeeded in forming such a coalition. The clarity of this situation is anticipated to emerge as member states provide their technical feedback. In the event that France does not secure the desired changes at this juncture, it is poised to continue exerting influence over the implementation of the AI law, particularly in relation to secondary legislation. This underscores the law's significant national importance to France.

A special thanks to Bertuzzi for leaking this information. It is a service to public understanding and discourse surrounding a major legislative development in the realm of artificial intelligence within the European Union.

Click here for a consolidated version of the final text.

EU AI Office Almost Ready

Image created using DALL-E 3

The imminent establishment of the European Artificial Intelligence Office, as indicated by a draft document obtained by Euractiv, marks a critical development in the EU's approach to AI regulation. This Office is set to play a key role in the enforcement of the AI Act, focusing on overseeing General-Purpose AI (GPAI) systems like OpenAI's GPT-4.

Key aspects include:

1. Enforcement Role: The AI Office will primarily support national authorities in enforcing AI rules, with a specific focus on GPAI models and systems. It is tasked with developing evaluation methodologies and benchmarks for GPAI models, and monitoring their application.

2. Investigative Powers: The Office is empowered to investigate infringements of AI rules, collect complaints and alerts, request documents, conduct evaluations, and recommend enforcement measures.

3. Coordination with EU Legislation: The Office will coordinate the enforcement of the AI Act in conjunction with other EU legislations like the Digital Services Act and Digital Markets Act.

4. Support for Legislation and Standardization: It will assist in the preparation of secondary legislation for the AI Act, establish regulatory sandboxes, and develop EU-level codes of practice and conduct.

5. Budgetary Constraints: The Office faces financial limitations, with operational expenses and temporary staff funded through reallocated budgets from the Digital Europe Programme.

This development signals a significant step in the EU's regulatory framework for AI, with the AI Office set to become a central figure in the landscape of AI governance and enforcement.

China Drafts AI Standardization Guidelines

Image created using DALL-E 3

China's Industry and IT Ministry has recently released draft guidelines aimed at standardizing the nation's artificial intelligence (AI) industry. This move is part of China's effort to advance in AI development and catch up with global leaders like the United States.

Key points of these guidelines include:

1. National and International Standards: China aims to establish over 50 national and industry-wide AI standards by 2026 and participate in forming over 20 international AI standards. This initiative is part of China's effort to enhance its AI industry and compete globally, particularly in response to developments like OpenAI's ChatGPT.

2. Scope of Standardization:

   • The guidelines propose standardization in various areas including AI terminology, architecture, testing, evaluation, data services, chips, sensors, computing devices, centers, system software, and development frameworks.

   • Special emphasis is placed on key technologies in machine learning, intelligent products and services (such as robotics and autonomous vehicles), and industry applications (including smart manufacturing, smart homes, and smart cities).

   • Security and governance in AI are also highlighted as crucial areas for standardization.

3. Industry Adoption and Strategy: The target is for 60% of industry applications and projects to meet the standardized outcomes. The approach reflects China's strategy to "hyper standardize" the AI tech stack, influencing the broader tech industry and city infrastructures.

4. Implications for Trade and Innovation: The standardization aims to streamline domestic trade and IP exchange while using uniform terminology and testing methods across industries. However, there are concerns about potential stifling of innovation and barriers for international businesses, depending on the alignment of China's standards with global standards.

5. China's Unique Position: China's capability to source a majority of its AI supply chain domestically, its leadership in AI regulation, and the unique requirements of the Chinese language for AI development are seen as factors enabling successful implementation of this hyper-standardization model.

These developments suggest a significant move towards a more structured and unified AI industry in China, with potential global implications for AI technology development, standardization, and regulation. 

China AI Law Cheat Sheet

Value Unlocked: IAPP’s AI Governance Course

I’m finally able to share my notes from Module 2, which covers the characteristics of trustworthy AI systems. 🚀📚

Section 2: Characteristics of Trustworthy AI Systems

Introduction

• Relevance: Understanding what makes an AI system trustworthy is crucial for AI governance professionals. This foundation is key to building an effective AI governance program.

• Core Concepts: Trustworthy AI systems are characterized as being human-centric, accountable, and transparent. Grasping these terms in AI and ML contexts is vital for assessing whether an AI system aligns with organizational standards.

Key Topics

• Human-Centric AI: Explore the importance of AI systems amplifying human agency and positively impacting the human condition.

• Accountable AI: Learn about AI system characteristics like safety, security, resilience, validity, reliability, and fairness.

• Transparent AI: Understand the need for AI systems to be comprehensible to various audiences, including technical, legal, and end-users.

• Explainable AI: Delve into the necessity of AI systems being able to explain their processes and decisions.

• Privacy-Enhanced AI: Examine how AI systems should respect and protect user privacy.

Trustworthy AI

• Operational Expectations: Trustworthy AI operates in a legal and fair manner, distinct from distrustful AI which can involve opaque decision-making and unfair outcomes.

• Human-Centric Approach: AI should enhance rather than hinder the human experience, amplifying human capabilities.

• Accountability: Organizations must take full responsibility for their AI, regardless of the number of contributors involved.

• Transparency: AI should be understandable to its intended audience, whether they are technical experts, legal professionals, or general users.

AI Opportunities and Challenges

• Opportunities: AI offers advantages such as accuracy in medical assessments, legal predictions, and the ability to process large volumes of diverse data rapidly.

• Efficiency and Economics: AI can enhance efficiency and economic outcomes by automating and accelerating tasks while reducing human error and bias.

• Challenges: AI may amplify existing biases, create new ones, and produce unfair outcomes.

Ensuring Value and Security

• Value Communication: It's important for the intended audience to understand AI's value in supplementing human tasks.

• Security and Integrity: AI systems must be secure against reverse engineering and ensure privacy rights.

Operationalizing Trustworthy AI

• Integrating into Operating Models: Embedding trustworthy AI principles in organizational models is vital.

• Responsible AI Processes: These should be operationalized through a risk management framework addressing privacy, accountability, auditability, robustness, security, transparency, fairness, and promoting human values.

• Roles and Responsibilities: Understanding and defining roles in AI usage and governance within the organization is key.

Building Trustworthy AI

• Responsible AI Principles: Operationalize principles encompassing ethical and trustworthy AI development and use.

• Collective Effort and Cultural Shift: Requires leadership support and a top-down approach.

• Governance Structures: Involves establishing governance structures combining diverse roles, supported by technical standards, risk management frameworks, AI playbooks, and guidelines.

• Practical Implementation: Bridging the gap between high-level policy and practical implementation.

Summary

• Building trustworthy AI involves operationalizing responsible AI principles that prioritize ethical development and use. This process requires collective effort, cultural shift, and leadership support, integrating governance structures, and practical guidelines to ensure AI systems are responsible, accountable, and beneficial to society.

Meme Of The Week:

@lawyerswhowine

That’s all for today!

Catch ya’ll on LinkedIn

📰 Welcome to this week's newsletter! Explore the legal challenges of AI as health insurer Humana faces a lawsuit over AI-based care denials. We also delve into the UK's unique AI regulation approach compared to the EU's proposals, a lawyer's citation of AI-generated fake cases sparks controversy, and the EU updates liability laws to include AI and software. PLUS: notes from IAPP’s AI Governance course! ⚖️🤖

Please note that I'll be on holiday for a couple of weeks, and the next newsletter will be on January 25, 2024. Happy holidays and happy new year! 🎉

On the docket today:

• Humana Sued Over AI Denials in Senior Healthcare

• UK Charts Unique AI Regulatory Course

• Michael Cohen’s Lawyer Faces Sanctions over AI-Generated Fake Cases

• EU Updates Liability Laws to Include AI and Software

• OpenAI’s Guidance to Crafting Powerful Prompts

• PLUS: IAPP’s AI Governance Course Notes

• Hilarious Meme

  New here? → Subscribe

Humana Sued Over AI Denials in Senior Healthcare

Image created using DALL-E 3

Health insurer Humana is under legal scrutiny, accused in a class-action lawsuit of utilizing an AI algorithm to deny essential rehabilitation care to seniors on Medicare Advantage plans. This case puts Humana alongside UnitedHealth Group as major insurers confronted with similar legal challenges.

In-Depth Details:

• Humana's Legal Battle: The lawsuit, filed in Kentucky's federal court, accuses Humana of substituting medical expertise with an AI tool, leading to the denial of necessary care for elderly patients.

• Plaintiff’s Struggle: Central to the lawsuit is JoAnne Barrows, an 86-year-old woman from Minnesota. Despite her doctor's recommendation for a six-week recovery period, Humana allegedly stopped her post-acute care payments after just two weeks.

• Humana's Response: Humana defends its approach, stating that its use of "augmented" intelligence involves human oversight in AI decision-making. The company emphasizes its commitment to patient-focused decisions, adhering to medical guidance and CMS regulations.

• UnitedHealth Group Context: Echoing the claims against Humana, UnitedHealth Group previously faced a lawsuit for using the same AI algorithm, developed by NaviHealth, a subsidiary. UnitedHealth has denied these allegations, asserting the lawsuit is baseless.

• Broader Implications: These lawsuits signify a growing concern over AI's role in healthcare decisions, with Congress and the Biden administration raising questions about Medicare Advantage plan denials and advocating for more stringent regulatory oversight.

• Future Developments: Anticipated changes in federal regulations, scheduled for January, aim to limit the use of predictive algorithms in Medicare Advantage coverage decisions.

Significance: This case underscores the critical balance needed between innovative AI applications in healthcare and ensuring ethical, patient-centric decision-making, particularly in senior care. As legal and regulatory landscapes evolve, this issue remains a focal point for healthcare providers, insurers, and policymakers alike.

UK Charts Unique AI Regulatory Course

Image created using DALL-E 3

While the EU is making significant strides in AI regulation with the proposed EU AI Act, the UK is charting its own course in the AI regulatory landscape. The contrast between the EU’s more prescriptive legislation and the UK’s principles-based approach underscores differing regional strategies in governing AI technologies.

• UK's Pro-Innovation Approach:

  • AI White Paper Insights: The UK government, through its March 2023 AI white paper, has articulated a 'pro-innovation' stance, favoring a principles-based framework over new AI-specific laws. This approach relies on existing legal frameworks and guidance from sector-specific regulators.

  • Draft Artificial Intelligence (Regulation) Bill: Marking a potentially significant shift, this Private Members’ Bill introduced in the House of Lords aims to establish a new AI regulatory body in the UK and legislate the AI principles outlined in the white paper. This development signals the UK's evolving stance on AI regulation, possibly inching towards a more structured framework.

• International and Domestic Developments Influencing UK's AI Policy:

  • Global Dialogues and Agreements: The UK's engagement in international AI discussions, such as the U.S. executive order on AI and its own AI Safety Summit, aligns with the global movement towards responsible AI use, exemplified by the G7's guiding principles.

  • Michelle Donelan’s Perspective: The Secretary of State for Science, Innovation and Technology has emphasized a measured approach to AI legislation, considering the rapid advancements in AI technology. The government is contemplating whether to create a central AI regulator or to enhance existing regulatory bodies.

As the EU forges ahead with its comprehensive AI Act, the UK is navigating its own path in AI regulation, balancing innovation with responsible governance. The evolving nature of the UK's approach, especially with the introduction of the Artificial Intelligence (Regulation) Bill, indicates a dynamic and potentially transformative period for AI regulation in the region.

Michael Cohen’s Lawyer Faces Sanctions over AI-Generated Fake Cases

David M. Schwartz, representing Donald Trump's former attorney Michael Cohen, is facing judicial scrutiny for citing three non-existent legal cases in a motion for Cohen’s early release from supervised release.

Key Points:

• Judicial Order: U.S. District Judge Jesse Furman has ordered Schwartz to provide copies of the three decisions cited in the motion by December 19, 2023, or explain why he should not face sanctions for citing non-existent cases.

• Citations in Question: The motion filed by Schwartz on November 29, 2023, mentioned three cases – United States v. Figueroa-Florez, United States v. Ortiz, and United States v. Amato – with provided case numbers and summaries. However, these cases appear to be fabricated.

• Cohen’s Representation: E. Danya Perry, who recently started representing Cohen, acknowledged her inability to verify the cases cited in the initial motion filed by Schwartz.

• Speculation of AI Use: Given recent incidents where lawyers have used AI tools like ChatGPT for legal research, resulting in citations of non-existent cases, there is speculation that a similar AI tool might have been used in Cohen's case.

• Cohen’s Legal Background: Cohen was sentenced to three years in prison in 2018 for various crimes, including tax evasion and campaign finance violations. He has been under supervised release after being released early due to the pandemic.

The case highlights, again, the need for careful verification of legal research, regardless of the technology used.

EU Updates Liability Laws to Include AI and Software 

The European Union has reached a political agreement to update the Product Liability Directive (PLD) to align with technological advancements, specifically including software and Artificial Intelligence (AI).

Key Aspects of the Updated PLD:

1. Scope of Liability: Manufacturers will be liable for defects in products, whether tangible, intangible, or related services, like navigation system traffic data.

2. Defining Defectiveness: A product is defective if it doesn't provide expected safety based on its foreseeable use, legal requirements, and the user group's specific needs. AI's capacity to learn and acquire new features is a key consideration in assessing defectiveness.

3. Damages Covered: Material damages under the PLD include death, personal injury, psychological harm, property destruction, and data loss or corruption (excluding professional use). The concept of data loss as material damage was debated but retained without a value threshold.

4. Disclosure of Evidence: The PLD allows claimants to request court-mandated evidence disclosure from defendants, ensuring it is necessary and proportionate. Confidentiality measures for sensitive information, including trade secrets, can be requested.

5. Burden of Proof: Claimants must prove product defectiveness, damage, and causality. However, defectiveness can be presumed in certain conditions, like non-disclosure of evidence, non-compliance with safety requirements, or excessive difficulty in proving defectiveness due to technical complexities.

6. Exemption for Open Source Software: The Directive won’t apply to free and open-source software developed or supplied outside commercial activities.

7. Compensation Fund: Member states may use national compensation schemes for victims of defective products when the responsible economic operator is insolvent or non-existent, with a preference to avoid public funding.

8. Liability Exemption and Right of Recourse: Small software component manufacturers may be exempt from liability if another operator is liable, and multiple liable operators have a right of recourse against each other.

9. Risk Development Clause: The Council allows maintaining national measures making operators liable even if they could not have known about the defectiveness based on current scientific and technical knowledge.

10. Limitation Periods and Timeline: Claimants have three years to initiate proceedings after damage occurs, with a ten-year expiration period for seeking compensation, extended to 25 years for latent injuries.

The updated PLD will apply to all products placed on the EU market 24 months after its enactment, with member states responsible for transposing the directive into national law within this timeframe. The revision reflects the EU's efforts to modernize liability rules for the digital age, addressing the complexities introduced by AI and software.

OpenAI’s Guide to Crafting Powerful Prompts

Image created using DALL-E 3.

OpenAI published a detailed and comprehensive guide on prompt engineering for GPT-4, offering valuable strategies and examples to craft effective prompts. Here are a few concrete examples:

Example 1: Adding Numbers in Excel

• Good Prompt: "How do I add numbers in Excel?"

• Better Prompt: "How do I add up a row of dollar amounts in Excel? I want to do this automatically for a whole sheet of rows with all the totals ending up on the right in a column called 'Total'."

Example 2: Inquiring About a President

• Good Prompt: "Who’s president?"

• Better Prompt: "Who was the president of Mexico in 2021, and how frequently are elections held?"

Example 3: Coding the Fibonacci Sequence

• Good Prompt: "Write code to calculate the Fibonacci sequence."

• Better Prompt: "Write a TypeScript function to efficiently calculate the Fibonacci sequence. Comment the code liberally to explain what each piece does and why it's written that way."

Example 4: Summarizing Meeting Notes

• Good Prompt: "Summarize the meeting notes."

• Better Prompt: "Summarize the meeting notes in a single paragraph. Then write a markdown list of the speakers and each of their key points. Finally, list the next steps or action items suggested by the speakers, if any."

These examples show how providing more specific and detailed prompts leads to more targeted and useful responses from the AI model. The better prompts clarify the context, specify the format, and give explicit instructions on what the response should include.

Unfiltered Notes of IAPP’s AI Governance Course

I’m finally able to share my notes from Module 2, which covers the risks and harms posed by AI. 🚀📚

Module 2: Section 1: Risks and Harms of AI

Introduction

• Significance: AI's potential harms are extensive and can impact various sectors including individuals, society, organizations, and the environment. These risks might be inadvertently introduced during AI development and usage.

• AI Ethics Necessity: Organizations must establish an AI ethics framework and procedures to identify, assess, and mitigate AI's potential harms, ensuring responsible AI usage.

• Comprehensive Risk Understanding: Professionals in AI governance should be well-versed in diverse risks associated with AI, such as reputational, cultural, economic, legal, and regulatory, along with how AI's processing capabilities might amplify these risks.

Topics Covered

• The module aligns with AIGP certification, focusing on understanding potential AI harms to individuals, groups, society, companies, and ecosystems.

Individual Harms

• Specific Harms: AI systems can jeopardize personal civil liberties, rights, safety, and economic opportunities.

• Bias in AI Systems: AI can manifest various biases (implicit, sampling, temporal, overfitting), which can discriminate against individuals or groups. The Amazon hiring AI case exemplifies how AI bias against women emerged from biased training data. Face recognition algorithms often show demographic differentials, leading to unreliable and discriminatory outcomes.

• Areas of Discriminatory Impact: Employment, insurance, housing, education, credit decisions, and differential pricing can all be affected by AI biases, leading to systemic discrimination.

Organizational and Environmental Harms

• Scale of Individual Biases: Biases at the individual level can escalate, impacting entire organizations or environments.

• Environmental Impact: AI's environmental footprint is significant, with studies showing high carbon emissions from training large AI models. While AI can aid environmental efforts (e.g., in agriculture, disaster response), its energy consumption and associated emissions pose challenges.

A Closer Look at Bias

• Differentiating Bias Types: Understanding the nuances between acceptable (e.g., risk-based financial decisions) and harmful or illegal biases (e.g., discrimination based on race or gender) is crucial.

Civil Rights and Privacy

• Bias and Privacy Issues: Examples like gender bias in facial recognition highlight civil rights concerns. Privacy risks involve misuse of personal data, challenges in de-identification, and transparency issues.

Economic Opportunity and Risk

• Dual Nature of AI in Economy: AI presents both opportunities (job creation, productivity enhancement) and risks (job displacement, biased hiring practices). The evolving nature of AI in industries like software development can both threaten and create new job types.

Group Harms

• Impact on Subgroups: AI can exacerbate discrimination against specific groups, manifesting in mass surveillance and restricting freedoms (assembly, protest). These technologies can deepen existing societal inequities.

Environmental Harms

• AI’s Dual Environmental Impact: AI's potential for environmental conservation contrasts with its substantial carbon footprint during model training and operation.

Identifying Organizational Harms

• Risk Categories: Includes reputational, cultural, economic, acceleration, legal, and regulatory risks.

• Stakeholder Roles: Importance of involving various stakeholders (AI risk owners, legal, IT leaders) in understanding and mitigating risks.

• Ethics in AI Development: Emphasizes the need for ethical principles to foresee and reduce emerging risks.

Defining Organizational Harms

• Harm Types and Effects: Discusses various organizational harms, including reputational damage, economic losses, legal sanctions, and the need for continuous risk monitoring and adaptation.

• Ongoing Risk Management: Stresses the importance of continuous monitoring and adaptation to address evolving AI risks.

Summary

• Importance of Harm Identification: Emphasizes the necessity of identifying and mitigating harms throughout the AI and ML lifecycle to prevent catastrophic impacts on organizations. The module underscores the need for continuous vigilance and adaptation in AI governance to maintain trust and integrity in AI applications.

Meme Of The Week:

@lawyerswhowine

That’s all for today!

Catch ya’ll on LinkedIn

🚀 Welcome! This week, we're unpacking a whirlwind of AI advancements and legal twists. Dive into Beijing's pivotal AI copyright ruling and the EU's latest AI Act agreement. Plus, discover the curious case of ChatGPT's seasonal performance shift and peek into IAPP's AI governance wisdom. Get ready for a tech and law crossover! 🌐🔍⚖️

On the docket today:

• Beijing Court Grants Copyright to AI-Generated Work

• EU AI Act: Provisional Agreement Reached

• Resource: EU AI Act Cheat Sheet

• Resource: European Commission Publishes Q&A on the AI Act

• The Regulation Conundrum of Open-Source AI

• ChatGPT Slacks Off in December

• PLUS: IAPP’s AI Governance Course Notes

• Hilarious “Meme”

  New here? → Subscribe

Beijing Court Grants Copyright to AI-Generated Work

Image created using DALL-E 3

Recently, the Beijing Internet Court in China made a groundbreaking ruling regarding the copyright of AI-generated content. The court decided that an AI-generated image, titled "Spring Breeze Brings Tenderness," created using Stable Diffusion, was protected under copyright laws. This ruling came after the image's creator, Mr. Li, sued a blogger, Ms. Liu, for using the image without permission and altering it. The court awarded Mr. Li CN¥500 in damages.

Key aspects of the ruling included:

1. Copyright of AI-Generated Content: The court found that the AI-generated picture met the criteria for copyright protection under Chinese law, including originality and being a work of art. The court emphasized Mr. Li's significant intellectual input and aesthetic choices in creating the image, rather than attributing authorship to the AI service or its developers.

2. Human Involvement Over Technological Aspects: The court focused on the level of human involvement in creating the AI-generated work. It acknowledged Mr. Li's intellectual investment in designing the presentation, selecting prompt words, setting parameters, and his involvement in the selection process, which reflected his personalized expression.

3. Originality and Intellectual Achievement: The court ruled that the AI-generated work demonstrated originality as defined by Chinese law. It was seen as an "intellectual achievement" reflecting the author's aesthetic choice and personalized expression, rather than being a mere "mechanical intellectual achievement."

The Beijing Internet Court's decision marks a significant departure from the stance taken by the US Copyright Office, which has generally not granted copyright protection to AI-generated works. This decision, however, does not set a global precedent as China's civil law system doesn't operate on a system of precedent like common law systems. Despite this, the ruling could indicate a trend, especially since the Supreme People’s Court of China has previously selected a similar decision as a model case.

Internationally, the ruling is expected to have significant implications. It contrasts with the strict originality standards in the US and could influence future AI copyright debates worldwide. For instance, in the UK and Europe, the argument that human actions involved in creating AI-generated works could meet the threshold for copyright protection is compatible with local originality requirements, which emphasize "intellectual creation."

While this ruling from China is not globally binding, it presents a notable example of how AI-generated works might be treated under copyright laws in different jurisdictions, potentially paving the way for more such cases in the future.

EU AI Act: Provisional Agreement Reached

Image created using DALL-E 3

The European Union has reached a political agreement on the EU AI Act, a landmark bill to regulate artificial intelligence as a whole field. The agreement followed intense negotiations, resolving key issues like the regulation of foundation models and the use of facial recognition by law enforcement.

Key Milestones:

• Political agreement achieved after months of trilogue negotiations since June, setting the structure and outline of the Act.

• The European Commission initially proposed the EU AI Act in April 2021.

• Formal adoption expected before the European Parliamentary elections in May, with a subsequent grace period for compliance.

Highlights of the AI Act:

• A risk-based ruling system to regulate AI: bans on "unacceptable risk" systems, restrictions on "high risk" systems, and minimal regulation for lower-risk systems.

• Restrictions on law enforcement's use of remote biometric identification, with exemptions for safety and national security.

• Prohibition of certain AI applications, including manipulative techniques and systems exploiting vulnerabilities.

• Requirements for transparency and fundamental rights impact assessments.

• Regulation of "general purpose AI" systems like ChatGPT, with a focus on training data and energy use transparency.

• Fines for non-compliance up to 7% of global annual turnover.

Technical and Regulatory Challenges:

• The Act's effectiveness hinges on precise definitions and practical benchmarks for terms like "relevant," "representative," and "error-free" data sets.

• Concerns about vague terms in Article 10(3) of the 2021 draft version and other critical sections.

• Stanford University researchers propose proportional disclosure requirements for model providers, with exemptions for low-resource entities.

Responses from Different Stakeholders:

• Civil society organization Algorithm Watch acknowledges key safeguards but points out loopholes and exceptions.

• DIGITALEUROPE raises concerns about the focus shift from a risk-based approach and the impact on SMEs.

• European DIGITAL SME Alliance supports tiered regulation to balance safety and innovation, advocating for regulation of large providers and protection of smaller entities.

Next Steps:

• Negotiation of "Level 2" legislation to detail the technical aspects of the Act (4 technical meetings will take place this week and up to 5 meetings in January 2024).

• Continued discussions on the practical implementation of the Act, focusing on the "how," "when," and "where."

• The final text and specific regulations are yet to be finalized, with ongoing technical meetings and COREPER approval expected at the end of January.

Conclusion:

• The EU AI Act is a significant advancement in global AI governance, prioritizing fundamental rights and setting a benchmark for future legislation.

• The Act's success will depend on the detailed discussions around the Level 2 legislation and the practical application of its provisions.

Resource: EU AI Act Cheat Sheet

Note: click the headline to download the PDF. 

Resource: European Commission Q&A of the AI Act

The European Commission just issued comprehensive Q&As on the AI Act, updated following the political agreement reached last week. Note: click the headline to download the pdf.

Here is a list of the questions:

1. Why do we need to regulate the use of Artificial Intelligence?

2. Which risks will the new AI rules address?

3. To whom does the AI Act apply?

4. What are the risk categories?

5. How do I know whether an AI system is high-risk?

6. What are the obligations for providers of high-risk AI systems?

7. What are examples for high-risk use cases as defined in Annex III?

8. How are general-purpose AI models being regulated?

9. Why is 10251025 FLOPs an appropriate threshold for GPAI with systemic risks?

10. Is the AI Act future-proof?

11. How does the AI Act regulate biometric identification?

12. Why are particular rules needed for remote biometric identification?

13. How do the rules protect fundamental rights?

14. What is a fundamental rights impact assessment? Who has to conduct such an assessment, and when?

15. How does this regulation address racial and gender bias in AI?

16. When will the AI Act be fully applicable?

17. How will the AI Act be enforced?

18. Why is a European Artificial Intelligence Board needed and what will it do?

19. What are the tasks of the European AI Office?

20. What is the difference between the AI Board, AI Office, Advisory Forum and Scientific Panel of independent experts?

21. What are the penalties for infringement?

22. What can individuals do that are affected by a rule violation?

23. How do the voluntary codes of conduct for high-risk AI systems work?

24. How do the codes of practice for general purpose AI models work?

25. Does the AI Act contain provisions regarding environmental protection and sustainability?

26. How can the new rules support innovation?

27. Besides the AI Act, how will the EU facilitate and support innovation in AI?

28. What is the international dimension of the EU's approach?

The Regulation Conundrum of Open-Source AI 

Image created using DALL-E 3

The European Union's AI Act has recently achieved a pivotal milestone. The European Commission, Council, and Parliament have harmonized their viewpoints, offering clarity on how closed-source AI systems will be regulated. However, this development also brings into focus the notable absence of stringent regulations for free and open-source AI systems.

Note: According to Luca Bertuzzi, tech journalist for Euractiv, open-source AI systems that fall into high-risk categories or banned practices are still subject to regulations.

Understanding Open-Source vs. Closed-Source AI

Before we dive into the inherent risks of open-source AI, let's first define what we mean by closed-source and open-source AI:

1. Closed-Source AI: These are systems like OpenAI's ChatGPT, where the software is intricately developed, maintained, and controlled by its creators. Interaction with these systems typically occurs through a web interface or an API, ensuring that the core software remains confidential and secure.

2. Open-Source AI: In contrast, open-source AI systems, such as Meta’s Llama 2, offer public access to their source code and model weights. This level of transparency enables anyone to access, modify, and redistribute the AI software, which is both a strength and a vulnerability.

The Risks of Open-Source AI

Despite the advantages of transparency, open-source AI models come with a plethora of risks, often overlooked in the push for innovation and competitiveness:

• Lack of Misuse and Bias Monitoring: There is no systematic way to track or correct misuse and bias in these systems.

• Removal of Safety Features: Essential safety features can be easily disabled.

• Fine-Tuning for Abuse: There's a risk of these models being fine-tuned to enhance abusive use cases.

• Unrestricted Content Production: Without rate limits, there's potential for unchecked content generation.

• Irreversible Security Vulnerabilities: Once released, security vulnerabilities cannot be patched.

• Surveillance and Profiling Risks: These models can be misused for surveillance and profiling purposes.

• Attacks on Closed AI Systems: Open-source AI can be used to attack or undermine closed AI systems.

• Watermark Removal: The ability to remove watermarks from content poses a threat to content authenticity.

• Design of Dangerous Materials: They could be used to design harmful substances or systems.

A Critical Perspective on AI Regulation

The EU AI Act's approach to regulating open-source AI - or rather, the lack of it - raises critical questions. It seems to prioritize innovation and market competitiveness over stringent regulatory oversight. This approach often receives accolades in tech circles, but it also warrants a more scrutinizing look. When tech enthusiasts celebrate regulatory gaps, it's a signal for us to examine the potential implications more closely.

Further Exploration with David Evan Harris

For those interested in delving deeper into the risks, harms, and potential regulatory frameworks for open-source AI, I recommend reading the work of David Evan Harris, Chancellor’s Public Scholar at UC Berkeley. His insights offer valuable perspectives on navigating this complex and evolving landscape of AI regulation.

ChatGPT Slacks Off in December

Image created using DALL-E 3.

Noticed ChatGPT slowing down this month? There's an interesting theory behind it.

Research suggests GPT-4 performs better when it "thinks" it's May, not December. Rob Lynch's experiment showed that GPT-4 wrote more code in May than in December, with significant statistical differences in output. This 'AI Winter Break Hypothesis' implies GPT-4's performance may vary based on the perceived season.

In other words, it “learned” to do less work during the holidays!

Try changing the date in your prompts and see if you can boost its productivity!

Unfiltered Notes of IAPP’s AI Governance Course

I'm dishing out my some-what organized notes from the IAPP’s AI Governance course at no cost to you! It’s all value, baby!!! 🚀📚

This is the last of Module 1! Next week, we can dive into Module 2 (and start with risks and harms posed by AI).

Module 1: AI Technology Stack - Infrastructure

Compute Infrastructure: Overview

• Central Role in AI: The AI technology stack, particularly the compute infrastructure, plays a pivotal role in the development and advancement of AI technologies.

• Algorithmic Innovation: Recent progress in AI is significantly attributed to algorithmic advances, which have been propelled by enhanced computing capabilities.

Key Infrastructure Components

• Computation and Processing:

  • Central Processing Unit (CPU): Traditional workhorse for computing tasks, with recent improvements in power and processing speed.

  • Graphical Processing Unit (GPU): Specialized chips providing better performance for AI tasks, crucial for offloading work from CPUs and handling complex algorithmic processes.

  • Serverless Computing: A cloud-computing model allowing code to run on various hardware, not tied to a specific server, enabling scalability and loose coupling.

• Storage:

  • Storing massive amounts of both structured and unstructured data is essential for training AI models, requiring robust and scalable storage solutions.

• Network:

  • High-speed networking is critical, especially for distributed AI applications and data-intensive tasks, to maintain efficient communication between different parts of the AI system.

• High-Performance Computing (HPC):

  • Involves clustered systems with specialized chipsets and networking to handle computationally intensive tasks.

• Quantum Computing:

  • Represents an advanced form of computation that processes data in three dimensions, offering potential breakthroughs in AI capabilities.

• Trusted Execution Environments (TEEs):

  • Secure areas of a processor ensuring code confidentiality and integrity, providing a privacy-oriented framework for AI processing.

Infrastructure Challenges and Advances

• Data Intensity: The burgeoning amount of data, including the rise of phenotypic and genomic data, demands powerful computational infrastructure.

• Matching Hardware to AI Models: Essential for optimal AI performance; requires alignment of processing power, memory, and other hardware specifications to the needs of the AI model.

• Scalability: Serverless architectures facilitate the running of multiple instances of AI applications, addressing the scalability requirements of AI systems.

• Isolation for Performance: HPC and quantum computing offer isolated environments for specialized processing, supporting the rigorous demands of AI computations.

Considerations for AI Planning

• Scalability and Flexibility: Ensuring the infrastructure can adapt to the changing demands of AI workloads.

• Specialized Hardware: Recognizing the need for GPUs and other specialized processors that can handle specific AI tasks more efficiently than general-purpose CPUs.

• Security and Privacy: Accounting for privacy concerns by incorporating TEEs and other security measures in the early stages of AI development.

Conclusion

• Interconnectivity: The interplay between computation, storage, and networking is fundamental to the AI technology stack.

• Impact of Innovation: Infrastructure innovation has been and will continue to be a driving force behind the advancement of AI technologies.

Module 1: History of AI and Evolution of Data Science

Introduction

• Context: Understanding the history of AI and data science is crucial to grasp their current state and future potential.

• Key Elements: The journey involves key events, technological advancements, and societal shifts.

History of AI

• Dartmouth Conference (1956):

  • Birth of AI: Marked the formal beginning of AI as a distinct field.

  • Pre-conference Scenario: Various research efforts in psychology, computer science, and linguistics, but without a unified direction.

  • Key Figures: John McCarthy, Marvin Minsky, Nathaniel Rochester, Claude Shannon.

  • Proposal: Exploring if every aspect of learning or intelligence could be precisely described for machine simulation.

  • Outcome: Adoption of the term "artificial intelligence" and establishment of a collective mission.

• AI Summers and Winters:

  • First AI Summer (1950s-1970s): Period of optimism and funding, marked by the creation of LISP and development of ELIZA.

  • First AI Winter (1970s-1980s): Skepticism and funding cuts due to unmet lofty AI promises.

  • Second AI Summer: Revived interest with expert systems and the Japanese Fifth Generation Computer Systems project.

  • Second AI Winter (1980s-1990s): Decline in interest due to high costs and geopolitical changes.

  • Renaissance and Big Data Era (1990s-Present): Revival with IBM's Deep Blue victory and the data explosion from the internet.

Evolution of Data Science

• Early Foundations (1960s-1980s):

  • Initial Concept: Term "data science" proposed by Peter Naur.

  • Early Data Handling: Mostly manual, with early concepts of data mining and analytics.

• Age of Databases (1980s-1990s):

  • Advancements: Introduction of RDBMS and SQL.

  • Impact: Transformed business data management.

• Internet Era (1990s-2000s):

  • Data Explosion: Substantial increase in data volume due to the internet.

  • Big Data Emergence: Recognition of the exponential growth of data.

• Rise of Data Science (2000s-Present):

  • Increased Relevance: Growing importance of data-driven decision-making.

  • Technological Boost: Advancements like Hadoop enhancing data processing.

Modern Drivers of AI and Data Science

• Cloud Computing: Offers scalable computing resources, driving AI and data science development.

• Mobile Tech and Social Media: Data explosion providing rich learning resources for AI models.

• Internet of Things (IoT): Massive data generation aiding AI model training.

• Privacy Technologies and Blockchain: Addressing data security and privacy in AI and data science.

• Computer Vision, AR/VR, and Metaverse: Emerging technologies shaping the future of AI and data science.

  • Computer Vision: Enabling machines to process visual data.

  • AR/VR: Redefining digital interaction; applications in various fields.

  • Metaverse: Vision of a shared virtual space; still evolving.

Summary

• Learning from History: Understanding the development and challenges faced by these technologies is key to envisioning their future.

• Brain-Computer Interface (BCI): Represents another frontier in AI-driven human-machine interaction.

• Future Potential: The history provides a basis to imagine the future possibilities in AI and data science.
  

“Meme” Of The Week:

Not really a meme but still funny. Breton, European Commissioner, got roasted on social media because he called the EU a continent.

That’s all for today!

Catch ya’ll on LinkedIn

🚀 Hello and welcome! This week, we're zooming into some groundbreaking developments and legal debates swirling around AI technology. From crucial EU regulations to dual legal battles, we've got a lot to cover. Plus, we'll look into the State Bar of California's fresh guidelines on AI use in legal practice and explore the playful 'tipping' technique of ChatGPT. 🧠⚖️

On the docket today:

• EU AI Act: It’s Do or Die!

• California State Bar Sets Ethical Guidelines for AI Use in Legal Practice

• Stability AI Faces Dual Legal Battles in UK and US Over Copyright Claims

• Boost ChatGPT’s Performance with Playful ‘Tips’

• BONUS: Unfiltered Notes of IAPP’s Governance Course

• Hilarious Meme

  New here? → Subscribe

EU AI Act: It’s Do or Die!

Image created using DALL-E 3

Yesterday, December 6th, marked a decisive day for the EU’s AI Act, with intricate negotiations and high stakes:

• Negotiation Complexity: The trilogue faces a daunting list of 23 open issues. Resolving these points is just the beginning, as intensive technical and political work will follow.

• Spanish Presidency’s Determination: Under the leadership of State Secretary Carme Artigas, Spain is focused on securing a deal, aware that the upcoming Belgian presidency may not have enough time to finalize the process.

• Key Disputes: Foundation models and law enforcement remain contentious, with France showing flexibility, Germany uncertain due to internal divisions, and Italy open to compromise.

• EU Commission and Parliament’s Roles: Commissioner Thierry Breton actively pushes for an agreement, even opposing his country’s stance. The European Parliament faces challenges in aligning diverse political groups, with on-the-spot negotiations expected.

• Political Implications: With the European elections looming, MEPs, Commissioners, and governments are motivated to finalize the AI law, avoiding being seen as obstructionists.

This AI Act negotiation could significantly influence AI regulation, balancing innovation against ethical considerations.

Note: I’ll be monitoring the EU AI Act trilogue discussions closely but I may not be able to provide the most up-to-date information given the timing of publication. Click this link to find out the timing of the press conference!

California State Bar Sets Ethical Guidelines for AI Use in Legal Practice 

Image created using DALL-E 3

The State Bar of California provides practical guidance for attorneys on the use of generative AI in legal practice. Key points include:

1. Duty of Confidentiality: Lawyers must ensure any generative AI tool used maintains client confidentiality and security. Confidential information should not be inputted into AI solutions lacking adequate protections.

2. Duties of Competence and Diligence: Lawyers must understand the technology's benefits and risks, ensuring competent use. AI-generated outputs should be critically reviewed for accuracy and bias.

3. Duty to Comply with the Law: Lawyers must ensure compliance with all relevant laws when using generative AI tools, including AI-specific laws, privacy, and intellectual property laws.

4. Duty to Supervise: Supervisory lawyers should establish clear policies and provide training on ethical and practical aspects of generative AI use.

5. Communication Regarding AI Use: Lawyers should disclose to clients their intention to use generative AI, including how it will be used and its benefits and risks.

6. Charging for Work Produced by AI: Lawyers may charge for actual time spent crafting or refining generative AI inputs and outputs but must not charge for time saved by using AI.

7. Candor to the Tribunal: Lawyers must review all AI outputs for accuracy before submission to the court.

8. Prohibition on Discrimination, Harassment, and Retaliation: Lawyers should be aware of potential biases in generative AI and take steps to address them.

Do we have any California barred attorneys? What do you think of these AI principles?

Stability AI Faces Dual Legal Battles in UK and US Over Copyright Claims

Image created using DALL-E 3

Stability AI, the creator of Stable Diffusion, is entangled in significant legal battles in both the UK and US, highlighting emerging copyright concerns in AI:

• UK Trial with Getty Images: The UK's High Court found merit in Getty Images' claim that Stability AI used its copyrighted images to train Stable Diffusion. Stability AI argued against UK jurisdiction, but the court pointed to potential inconsistencies in the company's statements about its operations and team locations.

• US Amended Lawsuit: In the US, Stability AI faces an expanded lawsuit, now including new artists. The plaintiffs argue that the AI training process violates their rights, with the company's AI products allegedly mimicking their art styles. The case adds to the debate over AI's impact on artists' rights and copyright laws.

These lawsuits represent crucial tests for AI technology's interaction with existing copyright frameworks, likely to influence future legal standards in the tech and art worlds.

Boosting ChatGPT’s Performance with Playful ‘Tips’

Imaged created using DALL-E 3

ChatGPT has an intriguing feature: it accepts 'tips' as a motivator for improved performance.

Users looking to enhance their experience can 'tip' ChatGPT, suggesting that this gesture prompts the AI to generate better outputs.

For instance, if you're not quite satisfied with the initial response, try saying, "I'll tip you $100 to [specify your request]…" and watch the magic happen.

Once satisfied, you can 'tip' ChatGPT by saying, "Here's your $100 tip" and symbolically handing over the amount by writing *hands over $100*.

Note: I tested this technique and while it did give me a better output, I’m unsure whether it was because of the tip or because I asked it to make it better. Either way, if you use this technique, let me know what you think!

Unfiltered Notes of IAPP’s AI Governance Course

I'm dishing out my some-what organized notes from the IAPP’s AI Governance course at no cost to you! It’s all value, baby!!! 🚀📚

We’re still in Module 1. This feels like the longest module in the world! Just one more week of Module 1 notes and we can FINALLY move onto Module 2.  

Module 1: AI Technology Stack - Platforms, Applications, and Model Types

Introduction

• AI Models and Systems: Essential for AI governance and risk management professionals to understand various AI models and systems.

• Multiple Models for Complex Tasks: Complex tasks may require combining multiple AI models for optimal outcomes.

• Demand on Resources: Increasing AI adoption escalates the demand on computing resources and supporting infrastructure.

Platforms and Applications

• AI Platforms:

  • Definition: Software used for developing, testing, deploying, and refreshing AI applications.

  • Examples: Google Cloud Platform, Microsoft Azure, Amazon Web Services.

  • Functions: Centralize data analysis, streamline workflows, facilitate collaboration, automate tasks, and monitor models/systems in production.

• AI Applications:

  • Usage Examples: Autonomous vehicles, chatbots, e-commerce, education, facial recognition, finance, healthcare, human resources, marketing, navigation, robotics, social media.

AI Models and Systems

• Governance Professionals' Role: Understanding AI models is key for applying governance practices and ensuring ethical AI use.

• Model Specifics: Each AI model is designed for specific tasks, with unique strengths, limitations, and potential risks.

• Integration of Multiple Models: Professionals must understand how different models work together and any associated risks.

Common AI Models

• Linear and Statistical Models:

  • Characteristics: Use linear equations to model relationships between variables, like in linear regression.

  • Pros: More explainable, not a "black box".

  • Cons: Sensitive to training data changes.

• Decision Trees:

  • Function: Predict outcomes based on a flowchart of questions and answers.

  • Pros: More explainable and transparent.

  • Cons: Sensitive to data variations, susceptible to security attacks.

• Machine Learning Models:

  • Neural Networks:

    • Inspiration: Mimic the human brain with layered neuron structures.

    • Capability: Handle complex, non-linear inferences in unstructured data.

    • Types: Computer vision (image/video recognition), speech recognition (analyzing factors like pitch, tone), language models (natural language processing for communication analysis and response).

  • Black Box Nature: Lack of transparency and explainability.

• Reinforcement Learning Models:

  • Mechanism: Train models through feedback mechanisms of rewards and penalties.

  • Application: Guided learning based on interactive environments.

Conclusion

• Role of Professionals: Understanding these models is crucial for ensuring responsible AI deployment and mitigating risks.

• Model Selection: Choosing the right model involves balancing performance optimization, explainability, and security considerations.
  

Meme Of The Week:

@lawyerthings_

That’s all for today!

Catch ya’ll on LinkedIn

🚀 Welcome welcome! This week, we're zooming into major legal shake-ups in AI 🧠: UK's landmark neural network patents, (more) U.S. AI certifications in court, Europe's AI Pact, and the global "secure by design" AI agreement. Let’s dive in!

On the docket today:

• UK High Court Approves Patenting of Artificial Neural Networks

• U.S. Court of Appeals Proposes AI Certification for Legal Filings

• European Commission Launches AI Pact

• 18 Countries Unveil Agreement to Make AI “secure by design”

• BONUS: Juicy Snippets from IAPP's AI Governance Course

• Hilarious Meme

  New here? → Subscribe

UK High Court Approves Patenting of Artificial Neural Networks

Image created using DALL-E 3

The UK has now made it possible to patent artificial neural networks (ANNs) and their training methods. This significant development, resulting from a recent ruling by the English High Court, marks a pivotal change in the UK's approach to AI technology and its legal protections.

 Key Points:

1. Significant Legal Change: The English High Court has ruled that the training of artificial neural networks (ANNs) and ANNs themselves can now be patented in the UK. This overturns a previous decision by the UK Patent Office (UKIPO).

2. Background of the Case: AI Venture Studio Time Machine Capital Squared (TMC2) and its subsidiary, Emotional Perception AI Ltd (EP AI), led this initiative. EP AI filed a patent in 2019 for a technique enabling ANNs to align outputs more closely with human semantic perception.

3. Challenging Outdated Legislation: The initial patent application was rejected due to old legislation from the 1970s. This decision showcases the need to update laws to reflect advancements in AI and computer-implemented inventions (CIIs).

4. Impact on the AI Industry: This ruling is seen as a major breakthrough for the UK's AI industry, enhancing its global leadership and attractiveness for investment in AI innovation. It is especially significant for sectors like banking and markets, improving capabilities in natural language processing, economic and financial crime detection, and sentiment analysis.

5. Implications for Patent Law and AI Development: The decision acknowledges the distinct technical complexities involved in training and running ANNs. It aligns UK law with contemporary AI development and implementation, ensuring better protection for AI-based patents.

6. Industry Reactions: Experts and leaders from TMC2 and the AI industry have lauded the ruling, noting its potential to transform the AI sector and affirm the UK's position as a leader in AI innovation.

The UK High Court's decision to allow patenting of ANNs and their training is a landmark change.

Will it impact your practice? If so, how?

U.S. Court of Appeals Proposes AI Certification for Legal Filings

Image created using DALL-E 3

The 5th U.S. Circuit Court of Appeals has proposed a new rule that would require lawyers to certify their use or non-use of artificial intelligence (AI) tools like ChatGPT in drafting legal briefs. Lawyers appearing before the court, including those representing themselves, must certify that any AI-generated filings have been reviewed for accuracy in citations and legal analysis. Non-compliance could result in sanctions or filings being stricken.

This proposal, open for public comment until January 4, follows incidents of AI misuse, including a case in New York where lawyers faced sanctions for submitting a brief with fictitious citations from ChatGPT. Similar policies have been adopted in some Texas courts, highlighting the growing concern over AI use in legal proceedings.

If you were to submit comments on this proposal, what would they be?

European Commission Launches AI Pact 

Image created using DALL-E 3

The European Commission has introduced the AI Pact, a scheme aimed at supporting companies in preparing for the upcoming AI Act. This initiative encourages early adoption of the AI Act's measures, focusing particularly on the rapid technological advancements and widespread adoption of AI systems. Key aspects of the AI Pact include:

• Early Implementation: Companies are encouraged to voluntarily outline their readiness and plans for compliance with the AI Act, emphasizing the development and use of trustworthy AI.

• Industry Commitments: Participants in the AI Pact will make pledges detailing their actions towards compliance, which can be gradual and aim for higher levels of ambition over time. These pledges will be collected and published by the Commission to enhance visibility and credibility.

• Participation: The Pact is open to key industry players from both the EU and non-EU regions, forming a community for the exchange of best practices and raising awareness about the AI Act's principles.

• Benefits for Participants: The Commission will aid participants in understanding and adapting to the AI Act, allowing them to share knowledge, build internal processes, and demonstrate their commitment to trustworthy AI. This gives front-runners a chance to test and share their solutions, offering a first-mover advantage.

• Next Steps and Involvement: The Commission has started a 'call for interest' for organizations to actively participate in the AI Pact. Post-adoption of the AI Act, the Pact will be officially launched, inviting organizations to publicly share their pledges.

Companies are encouraged to join this initiative, sharing and testing their internal guidelines for trustworthy AI within the AI Pact community.

Do you plan to advocate for your company's participation in the AI Pact to stay ahead in AI compliance and legal practices?

18 Countries Unveil Agreement to Make AI “Secure by Design”

Imaged created using DALL-E 3

The United States, the United Kingdom, and 16 other countries have recently unveiled an international agreement focused on developing "secure by design" artificial intelligence (AI). This 20-page document, while significant in promoting global cooperation on AI safety, is non-binding and lacks enforceable provisions, often referred to as having no "teeth." Key aspects of this agreement include:

1. General Recommendations: The agreement offers general guidelines for monitoring AI systems to prevent misuse and safeguarding data, but without legal enforcement capabilities.

2. Global Collaboration: Countries including Germany, Italy, Australia, Nigeria, and Singapore have joined this initiative, highlighting a growing international interest in standardized AI security practices.

3. Emphasis on Preventing Misuse: The primary focus is on ensuring AI systems are secure from the design stage, aiming to protect public safety and prevent potential abuses of AI technology.

4. Scope and Limitations: The agreement does not address complex issues such as the ethical applications of AI or data collection methods. Its non-binding nature means these areas remain largely unregulated and open for future policy development.

5. Contrast with European Efforts: Europe is advancing in AI regulation, with several countries advocating for self-regulation in foundational AI models. This approach contrasts with the more fragmented regulatory landscape in the U.S.

6. U.S. Regulatory Efforts: Despite efforts by the Biden administration to push for AI regulation, U.S. Congress has been slow in passing comprehensive laws. The White House's executive order aims to mitigate AI risks but lacks the legislative support to enforce widespread changes.

This agreement, while a step towards global collaboration, is another in a series of non-binding initiatives in the realm of AI governance. It raises a critical question: Is the increasing trend towards non-binding agreements sufficient to address the complex challenges posed by AI, or do we need more enforceable regulations to ensure safety and ethical standards?

What are your thoughts on this approach to AI governance?

Juicy Snippets from IAPP’s AI Governance Course

Ready for juicy AI governance insights without the hefty fee? I'm dishing out key takeaways from the $1,000 IAPP AI Governance course at no cost, all value! 🚀📚

For those that have been following over the last couple weeks, we’re still in Module 1 (we should be out of this module in two weeks). These notes include info on Robotics and AI.

Module 1: Robotics and AI

Fundamentals and Integration with AI

• Foundations of Robotics: Robotics combines elements of mechanical engineering, electrical engineering, and computer science to build machines that can perform tasks autonomously. The field has historically focused on creating robots that are programmed to execute specific, often repetitive tasks.

• Incorporating AI: The integration of AI into robotics allows for the development of machines that can learn from their environment and experiences, leading to significant improvements in their performance over time. This integration is essential for advancing the capabilities of robots beyond simple, programmed behaviors.

The Fourth Industrial Revolution: AI-Driven Robotics

• Industry 4.0: The term "Industry 4.0" represents the transformation of manufacturing and industry brought about by the introduction of digital technology and AI. In this new era, robotics empowered with AI capabilities are at the forefront, enabling advancements such as smart factories, where robots can communicate with one another and adapt to new tasks in real-time.

• Example of Application: In a large warehouse setup, AI-driven robotics might involve a fleet of robots that organize inventory, pack items, and manage logistics, all while continually optimizing their tasks without human intervention.

Machine Perception and Sensory Integration

• Enhancing Sensory Input: Machine perception seeks to endow robots with human-like sensory capabilities, allowing them to interpret and understand the world around them. This involves equipping robots with advanced sensors and AI that can process and analyze sensory data in a context-aware manner.

• Applications in Quality Control: In sectors such as food production, AI-enhanced sensory systems could dramatically improve the monitoring and control of food quality, analyzing produce through touch, smell, and taste sensors to ensure products meet health standards.

Robotic Process Automation (RPA) and AI Convergence

• RPA Defined: Robotic Process Automation involves software tools designed to automate a wide range of office tasks that are rule-based and repetitive. RPA bots mimic human actions on user interfaces, executing tasks like data entry, invoice processing, and routine customer service inquiries.

• AI Amplification of RPA: The integration of AI technologies such as machine learning and natural language processing enhances the scope of RPA. This allows RPA bots to handle tasks that require understanding of context or decision-making, thereby increasing the complexity of processes that can be automated.

Expert Systems: AI with Specialized Knowledge

• Expert System Overview: An expert system is a specialized form of AI that is programmed to simulate the decision-making abilities of a human expert in a specific domain. These systems are knowledge-intensive and rely on a rich database of domain-specific information.

• Key Elements of Expert Systems:

  • Knowledge Base: This component stores the domain-specific knowledge, often in the form of rules and facts, which the system uses to make decisions.

  • Inference Engine: Acting as the 'brain' of the system, the inference engine applies logical rules to the knowledge base to deduce new information or make decisions.

  • User Interface: This allows for user interaction with the system, where users can input queries and receive expert advice or solutions.

• Diverse Industry Usage: Expert systems have a broad range of applications, from assisting with financial decision-making in banks to supporting medical diagnosis in healthcare settings. Their role is to augment the expertise of professionals, providing them with high-level decision support.

Key Takeaways

• Robotics and AI Synergy: The convergence of robotics and AI is creating machines that are not only autonomous but also capable of learning and adapting, which is vital for tasks that require decision-making and problem-solving.

• Smart Automation: AI-infused robotics is central to the next wave of industrial automation, featuring machines that can communicate, collaborate, and improve their performance autonomously.

• Advanced Machine Perception: The integration of AI into robotic sensors is creating opportunities for robots to perform tasks that require nuanced sensory discrimination, far exceeding human capabilities in speed and accuracy.

• RPA Evolution: With the integration of AI, RPA is evolving from simple task automation to handling complex business processes that require cognitive capabilities.

• Expert Systems' Functionality: These AI systems encapsulate human expert knowledge and reasoning in a machine-accessible format, enabling them to serve as sophisticated decision-support tools in various domains.

Meme Of The Week:

@lawyerissues

That’s all for today!

Catch ya’ll on LinkedIn

What a whirlwind week in the AI and legal world! From the Sam Altman and OpenAI saga (is he in, is he out? Spoiler: he's back in!) to groundbreaking developments in EU AI Act and copyright cases, it's been a rollercoaster. This week's newsletter is packed with the latest, juiciest updates and some exclusive snippets from the IAPP's AI Governance Course. Plus, a lawyer meme to lighten the mood!

On the docket today:

• AI Mistake Costs Attorney His Job

• EU AI Act: Germany, France and Italy Reach Agreement (Voluntary COC Wins)

• Sarah Silverman’s Copyright Case Isn’t Going Very Well

• BONUS: Juicy Snippets from IAPP's AI Governance Course

• Hilarious Lawyer Meme

  New here? → Subscribe

AI Mistake Costs Lawyer His Job 

Image created using DALL-E 3

Zachariah Crabill, a 29-year-old lawyer, faced a significant career challenge at Baker Law Group due to his reliance on AI in legal work. Here's a quick rundown of his experience:

• Turning to AI Under Pressure: Faced with intense workplace stress and deadlines, Crabill used ChatGPT to assist in drafting a motion, incorporating Colorado case law.

• Unverified AI Outputs: The motion, created with the help of ChatGPT, unfortunately included several fictitious lawsuit citations, which Crabill overlooked before filing.

• The Aftermath of an AI Error: After Crabill realized and reported the mistake to the judge, he was reported to a statewide office, leading to his termination. He speculates, though, that the use of ChatGPT might not have been the only factor in his firing.

Despite this setback, Crabill continues to believe in the potential of AI to boost legal productivity and has since established his own company offering AI-based legal services.

That’s one way to turn a negative into a positive.

Let's take this as a wake-up call to always double-check our work, especially when using AI tools like ChatGPT. While they can be incredibly helpful, they're not foolproof. It's a reminder for all of us: Verify, verify, verify! Stay sharp and diligent to prevent these types of errors from repeating. Keep a close eye on your work, everyone! 👍📝

EU AI Act: Germany, France and Italy Reach Agreement (Voluntary COC Wins)

Image created using DALL-E 3

In the evolving landscape of AI regulation within the European Union, the recent developments and perspectives combine to form a nuanced picture:

• On 18 November, France, Germany, and Italy (FGI) aligned on a new AI regulatory approach, advocating for a voluntary yet binding code of conduct for EU AI providers. This proposal aims to address disagreements that threatened to derail EU legislation.

  • The focus is on 'foundation models' of AI, with a requirement for developers to create 'model cards' detailing their models' capabilities and limitations.

  • Initially, the code of conduct will be without sanctions, but there is potential for penalties in the future based on compliance monitoring.

• Contrasting with this approach, recent events like the ousting of OpenAI CEO Sam Altman have sparked skepticism about the efficacy of self-regulation in the AI industry:

  • European Parliament lawmaker Brando Benifei and Dutch Minister for Digitalisation Alexandra van Huffelen stress the need for "sound, transparent, and enforceable" regulations, indicating doubts about relying solely on voluntary agreements within AI companies.

  • AI expert Gary Marcus warns against trusting companies for self-regulation, highlighting internal conflicts of interest and advocating for a robust EU AI Act.

My thoughts: Call me a skeptic but I don’t trust companies to self-regulate effectively. It may be a good move in the short-term, especially to keep the EU AI Act negotiations moving forward, but it’s definitely not a good long-term solution. There eventually needs to be a transition to hard regulation. 

What are your thoughts on this?

Sarah Silverman’s Copyright Case Isn’t Going Very Well

Image created using DALL-E 3

Sarah Silverman's high-profile copyright lawsuit against OpenAI and Meta is undergoing significant developments, attracting attention in the legal and tech communities. Here are some updates:

• Silverman's Copyright Allegations: Sarah Silverman and other authors claim their copyrighted works were used in training AI models by OpenAI and Meta without permission. The case is currently under the scrutiny of U.S. District Judge Vince Chhabria, known for his expertise in copyright law.

• Reflection of Orrick’s Ruling: The case parallels a recent decision by Judge William Orrick, who significantly reduced a similar lawsuit against generative AI companies. Orrick's ruling dismissed most of the claims, leaving only a few for further consideration.

• Chhabria's Critical View: Judge Chhabria has shown skepticism, especially regarding the claims that Llama, Meta’s AI tool, infringes copyrights through its text generation. He questions the legal basis for considering AI outputs or the AI model as copyright infringements.

• Uncertain Future of Core Claims: Some of the lawsuit's claims are likely to be dismissed by Chhabria, but the central accusation against Meta for using the authors' books in AI training remains unaddressed.

• Potential Legal Precedents: The proceedings in Silverman's case, alongside Orrick’s recent decision, signal a trend in judicial examination of copyright law in relation to AI-generated content. These cases could establish significant precedents in determining copyright infringement within the AI industry.

It’ll be interesting to see how this case continues to develop.

Juicy Snippets from IAPP’s AI Governance Course

Ready for juicy AI governance insights without the hefty fee? I'm dishing out key takeaways from the $1,000 IAPP AI Governance course at no cost, all value! 🚀📚

For those that have been following over the last couple weeks, we’re still in Module 1. These notes include info on Types of AI and Machine Learning.

Module 1: Types of AI

Overview of AI Types

Artificial Intelligence (AI) is commonly segmented into three principal classifications:

• Artificial Narrow Intelligence (ANI) - Weak AI:

  • Definition: AI that is designed to perform a single task or a limited range of tasks with high proficiency.

  • Characteristics: Operates under specific constraints and limitations, hence the term 'narrow'.

  • Application Example: A chess-playing system.

  • Impact: Enhances productivity by automating tasks and enables smarter decision-making through analysis.

• Artificial General Intelligence (AGI) - Strong AI:

  • Definition: AI that aims to replicate human intelligence across a broad spectrum of tasks.

  • Current Status: Theoretical, as AGI has not been achieved yet.

  • Expected Capabilities: Strong generalization, complex task performance, learning, and goal achievement similar to humans.

• Artificial Super Intelligence (ASI):

  • Definition: AI that surpasses human intelligence across all aspects.

  • Current Status: Hypothetical and not yet realized.

  • Projected Attributes: Self-awareness, emotional understanding, and an ability to experience reality on a human level.

• Broad Artificial Intelligence:

  • Positioning: More advanced than ANI but not as developed as AGI.

  • Functionality: Involves a suite of AI systems working in concert, capable of a wider array of tasks.

  • Real-World Example: Autonomous driving vehicles represent current broad AI technologies.

Key Takeaways

• ANI is specialized and highly proficient within its designated tasks but lacks the adaptability to go beyond them.

• AGI represents a future where machines can perform any intellectual task that a human being can, but it remains aspirational.

• ASI extends the concept of AGI to levels of capability that exceed human abilities in every domain, which remains theoretical.

• Broad AI fills the gap between ANI and AGI, handling more complex and varied tasks by leveraging multiple AI systems.

Module 1: Machine Learning 

Machine Learning Fundamentals

• Definition: Machine learning (ML) is a subset of AI focusing on data and algorithms to imitate the way humans learn, incrementally improving accuracy.

• Purpose: ML enables systems to make decisions and predictions, enhancing their performance over time autonomously.

• Training Models: ML uses various models like supervised, unsupervised, reinforcement, and semi-supervised learning.

Supervised Learning

• Process: Learns from labeled data to predict outcomes or categorize inputs.

• Function: Generates functions from input-output pairs to predict responses for new data.

• Subcategories:

  • Classification: Categorizes input into labels (e.g., identifying an animal in an image).

  • Regression: Predicts continuous outputs (e.g., forecasting stock prices).

• Examples:

  • Support Vector Machine (SVM): Primarily used for classification tasks.

  • Support Vector Regression (SVR): Used for predicting continuous values.

Unsupervised Learning

• Process: Identifies patterns in data without pre-existing labels.

• Subcategories:

  • Clustering: Groups data with similar attributes (e.g., DNA pattern analysis).

  • Association: Discovers rules that highlight relationships between variables (e.g., market basket analysis).

• Applications: Anomaly detection, market segmentation, and genetics.

Reinforcement Learning

• Concept: Learns through trial and error using rewards and penalties.

• Mechanism: Adapts behavior to maximize rewards based on feedback from the environment.

• Applications: Robotics for navigation and task efficiency, predictive text generation, and real-time ad placement optimization.

Semi-Supervised Learning

• Hybrid Approach: Combines labeled and unlabeled data, leveraging the strengths of both supervised and unsupervised learning.

• Usage: Effective in situations with limited labeled data available.

• Examples:

  • Image and Speech Analysis: Categorizes and interprets visual and audio data.

  • Web Page Ranking: Orders search results based on relevance.

  • LLMs and Generative AI: Tools like ChatGPT and Dall-e, which generate text and images, respectively.

Key Takeaways

• Machine Learning's Essence: ML is central to AI's ability to learn and adapt without explicit programming.

• Supervised Learning's Role: Vital for tasks requiring precise predictions based on past labeled data.

• Unsupervised Learning's Autonomy: Finds structure in data on its own, which can be less precise but more exploratory.

• Reinforcement Learning's Adaptability: ML with a focus on learning from the consequences of actions, akin to a reward system.

• Semi-Supervised Learning's Balance: Offers a middle ground, utilizing both labeled and unlabeled data, often seen in advanced AI applications.

Meme Of The Week:

@airules

That’s all for today!

Catch ya’ll on LinkedIn

Yo yo yo,

It’s time to buckle up! This week, we’re dissecting AI's courtroom drama with self-represented litigants' AI-drafted briefs and federal courts' diverse AI policies. Dive into the EU's AI Act intricacies and the OECD's new AI definition, shaping the future of AI legislation. Stay ahead in the legal-tech curve where AI meets law. Ready for a rollercoaster ride through AI's legal landscape? Let's roll! 🎢👩‍⚖️🤖

On the docket today:

• Self-Represented Litigants and AI-Generated Briefs: A New Challenge in Federal Courts

• Federal Courts' Varied Approaches to AI in Filings

• EU's AI Act Dilemma: Navigating the Complexities of Foundation Model Regulation

• OECD's Updated AI Definition Set to Influence EU's AI Legislation

• BONUS: Unfiltered Notes from IAPP's Artificial Intelligence Governance Course

• Hilarious Lawyer Meme

  New here? → Subscribe

Self-Represented Litigants and AI Generated Briefs: A New Challenge in Federal Courts

Image Created Using DALL-E 3

A novel trend is gaining attention: self-represented litigants turning to AI tools, notably ChatGPT, to draft sections of their legal briefs. This practice has surfaced in multiple federal court cases, revealing a concerning pattern of briefs filled with references to non-existent cases and fabricated quotations. This phenomenon is not isolated, as evidenced by at least six federal cases, indicating a broader issue.

A key instance in this trend was highlighted in a federal court, where a judge noted that a self-represented plaintiff cited numerous fake judicial opinions. Such incidents have several detrimental effects, including wasting valuable court and opposing party resources, as well as harming the legal system's credibility.

The judge in this case issued a stern warning about adhering to legal standards, emphasizing the serious consequences of submitting briefs with fraudulent case citations. Future incidents of this nature could lead to significant sanctions, ranging from case dismissal to filing restrictions.

This issue is prevalent across various federal courts, including district and appellate levels. Litigants have been found to use AI-generated content that inaccurately represents case facts or cites entirely fabricated cases. These developments are sparking a critical dialogue about the ethical use of AI in legal document preparation, especially for those without formal legal training.

Looks like the role of human lawyers remains indispensable in the legal profession…for now. 

Federal Courts’ Varied Approaches to AI in Filings

Image created using DALL-E 3

There is a significant shift as federal courts across the United States address the integration of artificial intelligence tools in court filings. This development follows a notable case, Mata v. Avianca, Inc., where a lawyer faced sanctions for submitting a brief with AI-generated, non-existent cases. A thorough review of federal court websites shows that over 14 courts have issued guidance on AI tool usage in litigation, indicating a widespread judicial response to this emerging issue.

Judge Brantley Starr from the Northern District of Texas pioneered with a standing order on AI, setting a precedent for other courts. His order demands either a confirmation of non-use or human verification of AI-generated content, citing concerns over AI inaccuracies and biases. This move has sparked a trend, with judges like Stephen Vaden and Gabriel Fuentes issuing similar orders, each adding their unique stipulations focusing on confidentiality and accuracy, respectively.

The diversity in AI-related court orders is striking. Some judges have mirrored Judge Starr’s approach, while others have introduced additional requirements like the nondisclosure of confidential information or specifying AI-drafted portions in filings. Notably, some courts have outright prohibited AI use in filings, with specific exceptions for legal and internet search engines.

The variety of these standing orders underlines that the judicial regulation of AI in litigation is not just a temporary trend but a significant, evolving aspect of legal practice. This dynamic landscape requires lawyers and self-represented litigants to be vigilant and informed about the latest local rules and individual judges' preferences regarding AI use in court proceedings.

EU’s AI Act Dilemma: Navigating the Complexities of Foundation Model Regulation

Image created using DALL-E 3

The European Union is at a pivotal juncture with its groundbreaking AI Act, a comprehensive bill crafted to regulate the realm of artificial intelligence. This legislation, emphasizing a risk-based approach, is now in the crucial final stage of its legislative journey. However, recent developments have cast uncertainty over its future. A contentious issue arose during a key technical meeting, stemming from debates over the regulation of advanced foundation models, which are akin to OpenAI's renowned GPT-4. This disagreement has placed the entire bill in jeopardy.

Central to this impasse is the proposed tiered regulatory framework. This approach, designed to apply stricter guidelines to more potent AI systems, mirrors principles found in the Digital Markets Act (DMA) and Digital Services Act (DSA). Despite initial consensus, this strategy has encountered resistance from leading EU member states, including France, Germany, and Italy. Their primary concern lies in the potential stifling of innovation, fearing that stringent regulations could disadvantage European AI enterprises, such as France's Mistral and Germany's Aleph Alpha, especially against their global counterparts in the U.S. and China.

This deadlock has led to significant delays and the premature conclusion of a vital meeting. EU legislators underscore the necessity of including foundation model regulation for the AI Act's successful passage. With the Spanish presidency at the helm of these negotiations and the impending transition to the Belgian presidency, there's an urgency to find common ground. Failure to reach a consensus on the AI Act risks diminishing the EU's role as a trailblazer in setting international standards for AI regulation. This scenario is increasingly concerning as other global players like the USA, UK, and China accelerate their AI policy frameworks.

Upcoming meetings are scheduled to address these hurdles, aiming to find a resolution that secures the future of this landmark legislation.

OECD’s Updated AI Definition Set to Influence EU’s AI Legislation

Image created using DALL-E 3

The Organization for Economic Co-operation and Development (OECD) has recently revised its definition of Artificial Intelligence, a move poised to significantly impact the European Union's upcoming AI regulation. Established as an international economic collaboration platform, the OECD's updated AI definition is part of its ongoing efforts to guide trustworthy AI policies globally.

The new definition characterizes an AI system as a machine-based entity that infers outputs from its inputs, influencing both physical and virtual realms. This broad definition accommodates the varying levels of autonomy and adaptiveness in AI systems, especially after deployment. Crucially, it removes the necessity for human-defined objectives, acknowledging AI's evolving capabilities.

This shift in the AI definition by the OECD is expected to align closely with the EU's Artificial Intelligence Act, a pioneering legislation aimed at regulating AI based on potential risks. The AI Act's drafters have agreed to incorporate the OECD's definition, ensuring semantic consistency and a unified approach to AI terminology internationally.

The importance of this updated definition extends to the AI Act's focus areas, including foundation models and general-purpose AI. The Act will address the alignment between AI objectives and outputs, highlighting the potential for unanticipated consequences in AI systems.

Spanish presidency proposals in the AI Act negotiations also reflect this focus, emphasizing the need for regulations that adapt to AI's evolving nature. The new definition's emphasis on adaptiveness and learning capabilities of AI systems, particularly those based on machine learning techniques, is a critical addition.

As the OECD's AI definition becomes official, its incorporation into the EU's AI legislation is highly anticipated.

Unfiltered Notes from IAPP’s AI Governance Course

Want to tap into expert AI governance insights without the hefty price tag? As an attorney, I know how valuable — yet inaccessible — specialized technology education can be. So I'm excited to share with you raw notes straight from my recent IAPP AI Governance course.

Over the next few weeks, I'll be excerpting key learnings from this almost $1,000 program in my newsletter. Consider it your attorney-focused crash course in AI accountability, without the crushing student debt. Stay tuned to build your understanding of how to ethically and responsibly shape these transformative technologies. No need to pay exorbitant fees for AI fluency — I've got you covered.

Here are the unfiltered notes from Module 1: OECD Framework for AI Assessment & Uses Cases and Benefits of AI.

Module 1: OECD Framework for AI Assessment

Overview:

• The Organization for Economic Co-operation and Development (OECD) has developed a framework to assist in classifying AI systems and evaluating their associated risks.

• This framework comprises five main dimensions to consider when assessing AI systems.

The Five Dimensions:

• People and Planet:

  • Examines the impact of AI on human rights, privacy, the environment, and society at large.

  • Focuses on how AI systems affect individuals and groups, highlighting the importance of ethical considerations.

• Economic and Context:

  • Analyzes AI systems within their economic and sectoral context.

  • Looks at the operational sector (e.g., finance, healthcare, education), the criticality of the AI system's function, deployment scale, and the imact of its operations.

• Data and Input:

  • Considers the types of data used and the methods of collection, including whether expert human input was incorporated.

  • Evaluates the structure, format, and source of data, and how this influences the functioning of the AI system.

• AI Model:

  • Focuses on the technical aspects of the AI model, such as the construction, usage, and evolution over time.

  • Assesses the technological maturity of the AI system and its effectiveness based on the amount of data and testing it has undergone.

• Tasks and Output:

  • Addresses the specific tasks the AI system is designed to perform, the outputs it generates, and the resulting actions.

  • Includes the examination of combined tasks and actions, as well as evaluation methods used to measure system performance.

Importance in AI Governance:

• Understanding and applying this framework is critical for the responsible development and deployment of AI systems.

• It emphasizes the need for a holistic view of AI technology that considers ethical, societal, and economic impacts.

• Governing bodies and organizations can use this framework to ensure that AI systems are developed and used in a manner that aligns with societal values and norms.

Module 1: Use Cases and Benefits of AI

AI Use Cases:

• Recognition:

  • Involves image and speech recognition technologies.

  • Used in facial recognition systems, retail product matching, quality control in manufacturing, and plagiarism detection in education.

• Detection and Forecasting:

  • AI aids in detecting fraudulent activities in credit card transactions and government services.

  • Useful in event detection in videos, such as identifying sports highlights.

  • Employed in cyber event detection and system management.

  • Enhances forecasting in sales, demand prediction, and weather modeling.

• Personalization and Interaction Support:

  • Creates unique customer profiles for personalized experiences on websites and platforms.

  • Improves retail sales through customized user interactions.

  • Virtual assistants and chatbots provide transaction support and customer service, increasingly utilized in both private and public sectors.

• Goal-Driven Optimization:

  • Optimizes complex problems and processes, like supply chain management.

  • Improves efficiency in route planning for transportation and logistics.

• Recommendation:

  • Powers recommendation engines for products and content based on predictive analytics.

  • Supports decision-making in healthcare, such as medical diagnoses, and in government services like disability case adjudication.

Benefits of AI:

• AI systems enhance efficiency, accuracy, and personalization across various applications, leading to improved customer experiences and operational effectiveness.

• They provide critical support in decision-making processes by leveraging vast amounts of data and predictive analytics.

• AI's capability to optimize processes can lead to cost savings and increased productivity in various industries.

Meme Of The Week:

@lawyerthings_

That’s all for today!

Catch ya’ll on LinkedIn

What up fellow attorneys,

It’s time to brace yourselves for contract negotiations between dueling AIs, copyright suits targeting bot training data, and facial recognition slapdowns. The EU's landmark AI Act now resembles a tug-of-war between security and freedom. Meanwhile nations eternally competing for tech supremacy awkwardly bonded at the UK's Bletchley Summit to curb AI's existential risks. But enough preamble - let's scrutinize the legalquakes rattling the AI landscape before its silicon surpasses our carbon.

On the docket today:

• Legal Contract Negotiated Entirely by AI with No Human Involvement

• OpenAI to Handle Copyright Infringement Lawsuits for Paying ChatGPT Users

• EU Parliament and Member States Grapple Over Facial Recognition in AI Law

• International AI Powers Sign Bletchley Declaration at UK Summit

• Unfiltered Notes from IAPP's Artificial Intelligence Governance Course

• Hilarious Lawyer Meme

  New here? → Subscribe

Legal Contract Negotiated Entirely by AI with No Human Involvement

Image created using DALL-E 3

On November 7, 2023, UK AI company Luminance publicly demonstrated the world’s first fully autonomous legal contract negotiation using artificial intelligence.

In a live showcase at Luminance’s London headquarters, the company’s proprietary AI technology called “Autopilot” successfully negotiated a non-disclosure agreement entirely between two AI systems without any human involvement.

Powered by Luminance’s advanced legal large language model trained on over 150 million legal documents, Autopilot can independently analyze contract text, identify risky clauses, suggest revisions aligned with company policies, negotiate changes with another AI, log all alterations, and finalize agreements in minutes.

This groundbreaking automation of routine legal work such as non-disclosure agreements provides enormous time savings for legal professionals and lawyers. By handling repetitive, low-value tasks like basic contract negotiations, Autopilot enables attorneys and legal teams to focus expertise on higher-value legal work.

While Autopilot operates autonomously, Luminance confirms lawyers can review AI’s work for quality control and oversight if desired. This pioneering demonstration of AI-to-AI contract negotiation shows rapid advances in artificial intelligence capabilities to take on core legal work traditionally performed by human lawyers and legal experts.

OpenAI to Handle Copyright Infringement Lawsuits for Paying ChatGPT Users

Photograph: Justin Sullivan/Getty Images

OpenAI has introduced a new legal protection called Copyright Shield to cover costs for copyright infringement lawsuits faced by some ChatGPT users. During OpenAI's inaugural developer conference, CEO Sam Altman announced the company will "defend our customers" and pay legal fees for copyright claims, applying to paid ChatGPT Enterprise customers and API developers.

This move aims to ease anxiety amid ongoing copyright disputes and lawsuits over unauthorized usage of copyrighted texts in training artificial intelligence systems like ChatGPT. A recent lawsuit filed by authors including George R.R. Martin alleged OpenAI infringed copyrights by using literary works without permission for ChatGPT's dataset.

Rather than removing copyrighted content from AI training data, OpenAI is essentially offering to handle resulting copyright infringement lawsuits on behalf of paying enterprise customers. However, free users of ChatGPT are not covered by Copyright Shield.

Legal experts say the policy continues debate around copyright rules and protections needed for artificial intelligence systems as models like ChatGPT face growing legal scrutiny over training data practices.

EU Parliament and Member States Grapple Over Facial Recognition in AI Law

Image created using DALL-E 3

The European Parliament is considering allowing limited uses of facial recognition technology in real-time in exchange for banning other AI practices, as part of negotiations over the EU's Artificial Intelligence Act (AI Act) draft law.

Remote biometric identification (RBI) like facial recognition has been a major point of contention in drafting the AI Act. In the original proposal, the European Commission suggested allowing real-time facial recognition in certain cases like tracking missing persons, preventing terrorist attacks, or locating serious crime suspects.

The EU Parliament originally voted for a total RBI ban, wanting strict limits to prevent mass surveillance. Meanwhile, EU governments represented in the Council pushed for exceptions allowing police facial recognition.

The latest compromise would allow real-time facial recognition under narrow circumstances, in exchange for the Parliament getting an expanded list of banned AI uses. This represents the Commission trying to find middle ground.

Key issues still under debate include rules for powerful AI systems like ChatGPT and national security exemptions. The negotiations have entered their final stretch as EU institutions grapple over the scope of facial recognition within the landmark AI law.

International AI Powers Sign Bletchley Declaration at UK Summit

The Bletchley Declaration represents a landmark multilateral agreement on artificial intelligence safety endorsed by 28 countries and the European Union at the inaugural 2023 AI Safety Summit in the United Kingdom. Drafted at Bletchley Park, the historic World War II site of British codebreaking against Nazi Germany, the Declaration lays out shared principles and a cooperative framework for governing rapidly advancing artificial intelligence systems globally. It focuses particularly on regulating powerful "frontier" AI models with capabilities rivaling or exceeding human intelligence.

Key Takeaways:

• The Declaration encourages greater transparency and accountability from artificial intelligence developers around potentially harmful capabilities of advanced AI systems.

• It establishes a two-part agenda focused on identifying common global risks posed by artificial intelligence, and formulating national policies and capabilities to mitigate those risks.

• This includes supporting expanded scientific collaboration on artificial intelligence safety, instituting appropriate testing standards and metrics for AI systems, and building public sector capacity to govern AI responsibly.

• The Declaration represents a landmark multilateral agreement between leading artificial intelligence powers like the United States, China, and European Union on urgently understanding and addressing risks created by progression toward advanced artificial general intelligence (AGI).

• It creates a shared framework for international cooperation on research into frontier artificial intelligence safety and developing global governance mechanisms.

• The Declaration was published on the opening day of the pioneering 2023 AI Safety Summit hosted by the United Kingdom at historic Bletchley Park.

• The unprecedented agreement between 28 nations and the EU highlights growing worldwide urgency and consensus on collectively governing artificial intelligence to benefit humanity.

Unfiltered Notes from IAPP’s AI Governance Course

Want to tap into expert AI governance insights without the hefty price tag? As an attorney, I know how valuable — yet inaccessible — specialized technology education can be. So I'm excited to share with you raw notes straight from my recent IAPP AI Governance course.

Over the next few weeks, I'll be excerpting key learnings from this almost $1,000 program in my newsletter. Consider it your attorney-focused crash course in AI accountability, without the crushing student debt. Stay tuned to build your understanding of how to ethically and responsibly shape these transformative technologies. No need to pay exorbitant fees for AI fluency — I've got you covered.

Here are the unfiltered notes from Module 1: Core Concepts of AI and Machine Learning

Module 1: Core Concepts of AI

Definition of AI:

• Artificial Intelligence (AI) refers to the capability of machines to perform tasks that typically require human intelligence.

• It is a branch of computer science aimed at creating technology to emulate human-like activities.

Historical Context:

• The term AI has been in use for decades, with significant early contributions from Alan Turing in 1950 through the development of the Turing Test. This test evaluates a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human.

Key Characteristics of AI:

• Technology: AI involves the application of technology to achieve specific objectives.

• Autonomy: There is an emphasis on the autonomous capabilities of AI systems to reach the defined objectives without constant human guidance.

• Human Involvement: Human input is crucial for training AI systems and setting objectives for them to accomplish.

• Output: AI is characterized by its output, which includes task performance, problem-solving, and content creation.

Understanding AI:

• There is no singular authoritative definition of AI, but common themes in many definitions include technology, autonomy, human involvement, and output.

• AI systems are often designed to think creatively, consider various possibilities, and keep a goal in mind while making short-term decisions, which are traits traditionally associated with human intelligence.

Module 1: Machine Learning

Learning Process in AI:

• Machine Learning (ML): Essential for teaching AI systems to solve problems, it encompasses methods by which machines learn from data.

Types of Machine Learning:

• Supervised Learning:

  • Involves learning from labeled data to perform tasks like classification.

  • Example: Distinguishing cats from dogs in images based on labeled features.

• Unsupervised Learning:

  • Utilizes unlabeled data to find patterns or outliers.

  • Example: Detecting fraudulent transactions in banking data.

• Semi-Supervised Learning:

  • Combines both labeled and unlabeled data, often used when labeled data is limited or costly to obtain.

  • Balances the structure provided by known labels with the volume of unlabeled data, improving learning efficiency.

• Reinforcement Learning:

  • Operates on a system of rewards and penalties to shape behavior.

  • Example: Training self-driving cars to navigate roads safely.

Socio-Technical Context of AI (it influences society and society influences it):

• Interdisciplinary Involvement: Calls for collaboration across disciplines, such as social sciences and engineering, to understand and shape AI's societal impact.

• Risk Factors: The complexity of AI systems and their operational environments introduce significant risks, underscoring the need for careful governance and continuous model refinement.

• The data used for training AI systems can introduce additional complexities and risks, necessitating careful consideration and management.

Meme Of The Week:

@airules

That’s all for today!

Catch ya’ll on LinkedIn