Welcome to another edition of Horizon AI,

Meta recently debuted a new image generation model across its products, called Muse Image. Aside from the technical improvements, the launch has also stirred up privacy concerns, as the new AI image maker can generate images using photos from public Instagram accounts without their owners ever knowing.

In today's issue, we share how you can easily opt out of this.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• OpenAI Debuts New GPT-5.6 Models and ChatGPT Work Agent

• Meta Launches Muse Spark 1.1 for AI Coding

• AI Tutorial: How to Prevent Meta from Using Your Instagram Images in AI

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Debuts New GPT-5.6 Models and ChatGPT Work Agent

OpenAI announced the public release of GPT-5.6 Sol, Terra, and Luna, roughly two weeks after the AI company limited the rollout to a "small group of trusted partners" at the request of the U.S. government.

Details:

• The GPT-5.6 lineup includes Sol, Terra, and Luna, offering high-performance, balanced, and budget-friendly options for developers and businesses.

• OpenAI says the new models are more efficient and cost-effective than previous generations, with Sol delivering significant improvements in AI coding while using fewer tokens.

• There’s also a new "max" reasoning effort mode and an "ultra" mode that uses coordinated subagents to solve highly complex tasks.

• The company describes GPT-5.6 as its strongest cybersecurity model yet, supporting tasks such as threat modeling, secure code review, vulnerability remediation, and defensive security testing.

• OpenAI also introduced ChatGPT Work, a new productivity assistant for enterprise teams that helps create documents, spreadsheets, presentations, and other workplace content across desktop, web, and mobile.

GPT-5.6 is now available through ChatGPT, Codex, and the OpenAI API, adding another major release to an increasingly competitive AI model landscape.

Source

TOGETHER WITH LEVANTA

7 Proven External Traffic Strategies

Most eCommerce brands running external traffic aren't scaling — they're just spending.

Wrong channels, no real attribution, and at the end of the month, still no clear answer to the question that matters: what actually drove revenue?

The brands getting it right aren't necessarily spending more. They've just stopped guessing. They know which channels pull weight on Amazon listings, which ones bleed budget, and why affiliate and creator traffic outperforms on ROI when it's set up correctly.

Levanta put together a free playbook breaking down 7 proven external traffic strategies — where each one works, where it falls apart, and what it takes to scale without it becoming a second job.

Inside you'll see how top brands are driving millions in off-Amazon revenue and why most channels underdeliver when brands don't know what to look for.

If you're serious about growing outside of PPC, this is worth 5 minutes.

Grab the FREE playbook

META

Meta Launches Muse Spark 1.1 for AI Coding

Meta has released Muse Spark 1.1, the latest version of its AI coding model, expanding its efforts to compete with coding assistants from OpenAI and Anthropic.

Details:

• Muse Spark 1.1 is designed for agentic coding, allowing it to plan and execute multi-step development tasks, fix bugs, manage software workflows, and assist with large code migrations.

• Meta says the model performs well on tasks that require coordinating multiple tools and services, making it suitable for enterprise automation and software development workflows.

• The company is positioning Spark 1.1 as a low-cost alternative, with pricing of $1.25 per million input tokens and $4.25 per million output tokens.

• CEO Mark Zuckerberg described Spark as a model focused on agentic performance, tool use, and computer interaction, while also hinting that additional AI models are expected to follow.

The launch comes as competition in AI coding continues to intensify, with major AI companies rapidly releasing new models aimed at developers and enterprise users.

Source

AI Tutorial

How to Prevent Meta from Using Your Instagram Images in AI

Meta's new Muse Image came with a feature that lets users create AI images based on people's Instagram photos. The problem is: any adult with a public Instagram account is automatically opted in, and they won't even be notified when their photos are being used.

The easiest solution for those who want to opt out is to switch their public Instagram account to private.

But for those who want to keep their account public:

1. Head to your profile and tap the three horizontal lines in the top-right corner.

2. Scroll down to "Sharing and reuse."

3. Look for the option that says, "Allow people to use your content on Instagram with AI features on Meta."

4. Toggle the setting off for both posts and reels.

Note: The feature is currently available in the U.S., and Meta says it plans to expand it to Facebook and Messenger in the future.

AI Tools to check out

✨ Reve: Create, edit, and remix images. Combine natural-language edits with a drag-and-drop image editor.

🎮 Tesana: Build entire games and worlds by chatting with AI.

📝 Reline: AI notepad for meetings, lectures & calls. It captures the call right on your device then writes the summary, action items and follow-ups, and sends them where your team works.

🔍 Ora Directory: The agentic index of the web. Search agent-ready products by capability or domain

🛑 Harold: Stop bad data entering your business. This tool extracts, checks, and fixes your documents before they hit your systems.

TOGETHER WITH LEVANTA

Stop Spending. Start Scaling.

Most brands running external traffic aren't scaling — they're just spending. Levanta's free playbook breaks down 7 proven strategies: what works, what bleeds budget, and how top brands drive millions in off-Amazon revenue.

{{→ Grab the FREE playbook

AI Findings/Resources

👀 The inside story of how Claude Code went from an internal CLI to Anthropic's coding agent, told by researchers, engineers and early users who built it

🤔 How software engineers really feel about AI coding

👩‍🏫 ChatGPT's new voice mode is surprisingly good at tutoring

👉 The unexpected way AI helped me appreciate my body

🔐 How did the government decide OpenAI’s frontier model was safe to release?

The latest in AI and Tech

ByteDance launches Seedream 5.0 Pro image generation model

The new AI image model focuses on both image generation and advanced editing. It shows improvement in text rendering, layout accuracy, and design quality, while introducing features such as precision editing, layer separation, multilingual support, and image replacement tools. Its release comes ahead of the company's expected launch of the Seedance 2.5 AI video model later this month.

Elon Musk says he was “clearly wrong” about Anthropic, calling the company the current leader in AI

Elon Musk praised Anthropic's Mythos and Fable models, calling the company the current leader in AI and saying he would never cut off its access to SpaceX's infrastructure despite being a competitor.

Google will label AI-generated ads across its platforms

Google is introducing a new feature that lets users see whether an ad was created or edited using AI. The disclosure will appear in the "My Ad Center" panel across Google Search, YouTube, and Google Discover, under a new "How this ad was made" section.

OpenAI finds ~30% of tasks in SWE Bench Pro are broken

The company has withdrawn its endorsement of SWE-Bench Pro after finding that nearly 30% of the benchmark's coding tasks are flawed. According to the company, many tests are overly strict, ambiguous, or misleading because they were adapted from real software projects rather than designed as standardized AI evaluations.

Mistral launches first AI model for robot navigation

The French AI startup has unveiled Robostral Navigate, its first AI model for robot navigation, capable of guiding wheeled, legged, and flying robots using only a single RGB camera. Trained entirely in simulated environments, the 8B-parameter model achieved a 79.4% success rate on the R2R-CE benchmark, outperforming several systems that rely on multiple cameras or depth sensors.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

OpenAI is rolling out a new voice experience for ChatGPT that makes conversations feel much more natural. In today's issue, we break down what's new and share a few things you can try with the upgraded experience.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• OpenAI Launches GPT-Live to Make ChatGPT Voice Conversations Feel More Natural

• SpaceXAI Launches Grok 4.5, Calling It an "Opus-Class" AI Model

• AI Tutorial: 5 Things to Try with ChatGPT's New Voice Mode

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Launches GPT-Live to Make ChatGPT Voice Conversations Feel More Natural

OpenAI has introduced GPT-Live-1 and GPT-Live-1 mini, two new voice models designed to make conversations with ChatGPT feel more natural, responsive, and capable of handling longer interactions.

Details:

• The new models use full-duplex voice technology, allowing them to listen and speak at the same time. This enables more natural interruptions, smoother turn-taking, and features such as real-time translation.

• Unlike the previous voice system, the new models can connect to OpenAI's latest AI models, like GPT-5.5, for tasks like reasoning, search, and agent-based workflows while continuing the conversation without interruption.

• OpenAI says the models are designed for longer voice sessions, with the ability to stay silent, listen to ongoing conversations, and respond only when needed while maintaining context.

• The updated voice experience can also display visual information when appropriate. It will supplement conversations about the weather, stocks, and sports with AI-generated visuals showing relevant information.

• GPT-Live-1 is rolling out across iOS, Android, and web, powering ChatGPT Voice for Go, Plus, and Pro subscribers, while a smaller, more efficient GPT-Live-1 mini model will be the default for free users.

OpenAI notes that in addition to being its "smartest voice model" yet, it includes built-in safeguards to steer conversations away from harmful responses or end chats entirely in "higher-risk" situations, offers "expert-vetted crisis helpline support" for conversations about self-harm, and provides "age-appropriate" responses for teens.

Source

TOGETHER WITH LEVANTA

The Free Playbook Behind Millions in Off-Amazon Revenue

Most eCommerce brands running external traffic aren't scaling — they're just spending.

Wrong channels, no real attribution, and at the end of the month, still no clear answer to the only question that matters: what actually moved your BSR?

The brands getting it right aren't necessarily spending more. They've just stopped guessing. They know which channels pull weight on Amazon listings, which ones look good in a dashboard but bleed budget, and why creator traffic consistently outperforms paid social on ROI when it's set up correctly.

Levanta put together a free playbook breaking down 7 proven external traffic strategies. Inside you'll see how top brands are driving millions in off-Amazon revenue and why most channels underdeliver when brands don't know what to look for before they start spending.

If you're serious about growing outside of PPC, this is worth 5 minutes.

Grab the FREE playbook

SPACEXAI

SpaceXAI Launches Grok 4.5, Calling It an "Opus-Class" AI Model

SpaceXAI has released Grok 4.5, its latest AI model, positioning it as a faster and more cost-efficient option for coding, research, writing, and other knowledge-intensive tasks.

Details:

• The company claims the model delivers twice the token efficiency of leading competitors, which could reduce inference costs while maintaining strong performance.

• Elon Musk described Grok 4.5 as an "Opus-class model," saying its capabilities are roughly comparable to Anthropic's Opus 4.7 while offering faster responses and lower operating costs.

• According to SpaceXAI, Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens, making it significantly cheaper than several frontier AI models.

• The company will also be offering free Grok 4.5 usage "for a limited time" in Grok Build and Cursor.

The company's shared benchmarks show Grok's competitiveness with other top models, and although it falls just short of best-in-class overall, they claim that "the combination of capability, faster speed, and lower cost" will be what makes it competitive as the company prepares to release a 2T model "next month."

Source

AI Tutorial

5 Things to Try with ChatGPT's New Voice Mode

Check if GPT-Live has arrived in your account:

• Mobile: Tap the menu (top left) → settings cog (Android) or profile avatar (iOS) → Voice > Model

• Web: Click your profile avatar (bottom left) → Settings > Voice > Model

Then start a conversation and try these:

1. One of the biggest upgrades is "duplex" functionality: try to keep talking even after ChatGPT has started answering, and it should now be able to keep up.

2. Try interrupting ChatGPT mid-response and changing your question. It should adapt naturally without restarting the conversation.

3. Ask ChatGPT to translate what you're saying into another language in real time. Speak sentences in English, and ChatGPT will translate while you talk.

4. Tap the sliders icon to switch between Instant, Medium, and High intelligence modes. Use Instant for the fastest answers, High for the best answers, and Medium as a compromise.

5. Finally, ask about the weather, sports, or places. The new voice mode can display visual cards alongside its spoken responses, making conversations more interactive.

AI Tools to check out

🔍 AnySearch: A privacy-first search infrastructure for AI Agents, with anonymous access, smart routing, structured output, and native API, MCP.

🚀 RankAI: Grow your website on autopilot. It reads your live site, ships on-brand pages that rank and convert, and keeps improving them.

⭐ Branda: Turn a one-sentence idea into a full brand — strategy, name, logo, palette, type, and a brand kit.

🛠️ Timbal AI: Build AI agents, workflows, and apps in one stack.

👀 ExploreYC: Explore Y Combinator companies with powerful search, interactive maps, AI-powered company intelligence, hiring insights, funding data, and batch analytics.

TOGETHER WITH LEVANTA

7 Strategies. Zero Guesswork.

The brands winning on Amazon stopped guessing. They know which external channels pull weight — and which ones don't. Levanta's free playbook breaks down 7 strategies driving millions in off-Amazon revenue.

→ Grab the FREE playbook

AI Findings/Resources

🤔 Why the rise of open source AI isn’t hurting Anthropic … yet

🔇 Research lab Aleph built an AI system that can turn tongue movements into text without using sound

🚨 Meta says it’s facing $1.4T in penalties in teen mental health case — a sum equal to tech giant’s valuation

😱 AI faces trusted more than faces of real people, warn researchers

👀 The people who will thrive in the AI age

The latest in AI and Tech

Google Photos adds a new AI 'Video Remix' tool

Google has introduced Video Remix for Google Photos, adding AI-powered editing features such as cinematic relighting, custom backgrounds, and artistic effects like watercolor and sketchbook styles. Powered by Gemini Omni, the feature is rolling out to Google AI Plus, Pro, and Ultra subscribers starting today.

Anthropic is testing a new Reflection tool

The new tool helps Claude users see when they use the chatbot, what they use it for, and whether that usage lines up with their goals. It is rolling out to Free, Pro, and Max users with Memory enabled.

Google’s SynthID watermark just debunked its first high-profile deepfake

Earlier this week, a picture circulated online on platforms like Reddit and X that seemed to show Kentucky Senator Mitch McConnell covered in tubes in a hospital bed. On Wednesday, Snopes, a fact-checking website, debunked the image, confirming that it contained the SynthID watermark that Google designed to identify AI-generated pictures.

Ollama raises $65M for its open-model AI platform

Open-source AI developer tool Ollama has raised $65 million in a Series B funding round, bringing its total funding to $88 million. The company says its platform is now used by nearly 9 million developers each month, helping them run open AI models locally and in the cloud.

Lovable in talks to double valuation

Swedish AI startup Lovable is reportedly in talks to raise $300 million at a $13.2 billion valuation, doubling its valuation from last December. The company recently reached a $500 million annualized revenue run rate and counts enterprises such as Workday, Asana, and Nvidia among its customers.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

One of the biggest reasons AI gives disappointing answers is that it's often forced to guess what you mean. Instead of asking for clarification, ChatGPT will often just make assumptions.

In today's issue, we share a simple prompt you can add that encourages the AI to ask questions first, so you get more accurate and useful responses.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Meta Launches Image Generation Model with Coding, Search Capabilities

• China Reportedly Considers Export Restrictions on Advanced AI Models

• AI Tutorial: This One Sentence Can Instantly Improve Your ChatGPT Responses

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

META

Meta Launches Image Generation Model with Coding, Search Capabilities

Meta has introduced Muse Image, the first image generation model built by the company's Superintelligence Labs.

Details:

• Muse Image uses an AI agent approach, calling tools to produce more accurate results and refine its own outputs.

• Those tools include writing and running code to generate diagrams, scannable QR codes, animated GIFs, websites, or interactive games, as well as a web search feature that grounds images in current facts and real-world references, improving accuracy on knowledge-heavy prompts

• On Image Arena, it ranks second in human preference scores for text-to-image generation and for both single- and multi-image editing, trailing OpenAI's GPT Image 2 in each category but beating models like Nano Banana and Grok Imagine.

• One of its new features allows users to reference public Instagram accounts in prompts, enabling the model to generate new images using publicly available photos from those profiles. Users can opt out by changing their Instagram settings.

Muse Image is available through the Meta AI app, Instagram Stories, WhatsApp, and meta.ai, with additional integrations planned for other Meta products. The company also announced that a new 'Muse Video' model is in development.

Source

TOGETHER WITH LEVANTA

7 Proven External Traffic Strategies

Most eCommerce brands running external traffic aren't scaling — they're just spending.

Wrong channels, no real attribution, and at the end of the month, still no clear answer to the question that matters: what actually drove revenue?

The brands getting it right aren't necessarily spending more. They've just stopped guessing. They know which channels pull weight on Amazon listings, which ones bleed budget, and why affiliate and creator traffic outperforms on ROI when it's set up correctly.

Levanta put together a free playbook breaking down 7 proven external traffic strategies — where each one works, where it falls apart, and what it takes to scale without it becoming a second job.

Inside you'll see how top brands are driving millions in off-Amazon revenue and why most channels underdeliver when brands don't know what to look for.

If you're serious about growing outside of PPC, this is worth 5 minutes.

Grab the FREE playbook

AI RACE

China Reportedly Considers Export Restrictions on Advanced AI Models

China is reportedly considering new restrictions on access to its most advanced AI models, a move that could reshape global AI competition and make it harder for companies outside the country to use leading Chinese models.

Details:

• According to Reuters, Chinese officials have held discussions with major AI companies, including Alibaba, ByteDance, and Z.ai, about limiting foreign access to advanced and future AI models.

• The proposals reportedly include tighter controls on frontier AI systems, possible security reviews for advanced models, and new rules governing overseas investment in Chinese AI startups.

• The discussions come as Chinese models such as Qwen, Doubao, and GLM-5.2 have gained international attention for offering strong performance at lower costs.

The reported plans mirror recent U.S. restrictions on Anthropic's latest advanced AI models, highlighting how both countries are increasingly treating frontier AI as a strategic national asset.

Source

AI Tutorial

This One Sentence Can Instantly Improve Your ChatGPT Responses

1. Go to ChatGPT and write your normal prompts as you usually would.

2. Add this sentence to the end:

Before answering, ask up to three clarifying questions only if they're necessary to avoid making incorrect assumptions. Otherwise, answer immediately and mention any assumptions you made.

3. If the AI needs more information, answer its clarifying questions before it generates the final response. If it already has enough context, it'll answer immediately while explaining any assumptions it made.

You can also try this in Claude, Gemini, or another AI chatbot.

This prompt works best for tasks where there are multiple possible answers, such as writing emails, planning trips, brainstorming ideas, creating workout plans, coding projects, or making important decisions.

AI Tools to check out

🔥 ZCode: The official development environment for GLM-5.2. Combines the best AI agents with your existing tools so you can plan, code, review, and deploy without friction.

🔮 Futuresearch: Forecasting backed with a public track record on stocks, prediction markets, public benchmarks, and forecasting tournaments.

🏴‍☠️ The Hugging bay: The Pirate Bay for open LLMs — model weights downloadable via torrents.

🦾 Jira: A powerful project management tool that helps teams plan, organize, and automate their work.

💼 Badge: AI agents collect peer reviews to generate proof of work.

AI Findings/Resources

😅 Someone made a song called “Claude’s Plan” inspired by Drake’s God’s Plan

📙 The only AI glossary you’ll need this year

👀 Anthropic shares a strategy to offset Fable 5's high cost, suggesting its use as an 'advisor' that provides guidance to a faster, lower-cost executor model, which handles the majority of the work

🔨 Build your own AI agent from scratch using the Vercel AI SDK, Composio, Supermemory, Telegram, and Vercel Cron

🚨 Chinese AI models are gaining ground with U.S. companies as OpenAI, Anthropic costs surge

The latest in AI and Tech

OpenAI gets permission to roll out GPT-5.6 to the public on July 9

OpenAI has received approval from the Trump administration to publicly launch its GPT-5.6 model family on Thursday. The lineup includes Sol, the flagship model, Terra for high-volume workloads, and Luna, a faster, lower-cost model for everyday tasks.

Meta tests AI glasses that can continuously record daily life

Meta is reportedly testing AI-powered smart glasses with an always-on 'Super Sensing' mode that continuously captures audio and images throughout the day, allowing users to ask AI about things they've seen or heard. The prototype has sparked internal privacy concerns, and although Meta has not confirmed the project, it said it remains committed to privacy-focused technology.

MiniMax plans to open source 2.7 trillion parameter AI model

Chinese AI startup MiniMax is reportedly preparing to release a 2.7 trillion parameter large language model as open source later this year. The model, internally known as M3 Pro, would be the largest Chinese AI model to date and could launch as early as the third quarter.

OpenAI's chief futurist departs after nine years

OpenAI's chief futurist, Joshua Achiam, has announced he is leaving the company after nine years. In a post on X, Achiam said there was no single reason for his departure, adding that he believes he can continue contributing to the safe development of AGI from outside a frontier AI lab.

Google DeepMind expands Gemini API managed agents

Google DeepMind has introduced four new capabilities for Managed Agents in the Gemini API, including background execution, support for remote Model Context Protocol (MCP) servers, custom function integration, and credential refresh without resetting the agent's state.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

Even the best AI models still get things wrong, and a confident answer isn't always a correct one. Instead of relying on a single chatbot, you can compare responses across multiple leading AI models before using the information.

In today's issue, we'll show you a free tool that automatically checks your prompts across ChatGPT, Claude, Gemini, and other AI models, helping you spot differences, catch mistakes, and get more reliable answers without switching tabs.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Anthropic Unveils Claude's Hidden "J-Space" Internal Workspace

• Tencent Releases Hy3 Under Apache 2.0 License

• AI Tutorial: How to Get More Accurate Answers from ChatGPT, Claude, and Gemini

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

ANTHROPIC

Anthropic Unveils Claude's Hidden "J-Space" Internal Workspace

Anthropic has revealed a newly discovered internal mechanism inside its Claude models, called J-Space, offering researchers a closer look at how AI systems process information beyond the reasoning they show users.

Details:

• J-Space is an internal computational workspace that operates separately from the model's visible reasoning, allowing Claude to process concepts that never appear in its final response.

• In one demonstration, Claude was asked to copy a sentence while simultaneously thinking about the Golden Gate Bridge. Although the output only repeated the sentence, researchers observed internal activations related to concepts like "bridge" and "California."

• The researchers note that the J-Space wasn't designed or programmed by them, but instead "emerged on its own" during Claude's training process.

While the research discusses concepts related to consciousness, Anthropic emphasizes that the findings do not suggest Claude is conscious. Instead, the company frames J-Space as an important step toward opening the AI "black box," with the long-term goal of making advanced AI systems safer, more understandable, and easier to monitor.

Source

TOGETHER WITH LEVANTA

The Free Playbook Behind Millions in Off-Amazon Revenue

Most eCommerce brands running external traffic aren't scaling — they're just spending.

Wrong channels, no real attribution, and at the end of the month, still no clear answer to the only question that matters: what actually moved your BSR?

The brands getting it right aren't necessarily spending more. They've just stopped guessing. They know which channels pull weight on Amazon listings, which ones look good in a dashboard but bleed budget, and why creator traffic consistently outperforms paid social on ROI when it's set up correctly.

Levanta put together a free playbook breaking down 7 proven external traffic strategies. Inside you'll see how top brands are driving millions in off-Amazon revenue and why most channels underdeliver when brands don't know what to look for before they start spending.

If you're serious about growing outside of PPC, this is worth 5 minutes.

Grab the FREE playbook

TENCENT

Tencent Releases Hy3 Under Apache 2.0 License

Tencent has officially released its new AI model, Hy3, an open-source model the company claims matches the performance of models two to five times its size.

Details:

• Hy3 is a 295-billion-parameter mixture-of-experts (MoE) model with 21 billion active parameters, designed to deliver strong performance while requiring significantly less compute than larger frontier models.

• Tencent says Hy3 performs particularly well on agentic search, tool use, long-context retrieval, and AI workflows, outperforming many competing open models in these areas.

• While Hy3 trails GLM-5.2 on coding benchmarks, Tencent positions it as a more efficient alternative, offering competitive capabilities with less than half the total parameters (295B vs. 744B, with 21B vs. 40B active).

• The company also reports major improvements in reliability, including lower hallucination rates, fewer reasoning errors, and stronger multi-turn conversations compared to the preview version released earlier this year.

The company is releasing Hy3 under the permissive Apache 2.0 license on Hugging Face, ModelScope, and GitHub.

Source

AI Tutorial

In partnership with

How to Get More Accurate Answers from ChatGPT, Claude, and Gemini

You attach AI answers to real work every day. On Artificial Analysis's AA-Omniscience accuracy test, the top-ranked model, Claude Fable 5, still lands at just 61%.

Fluent does not mean correct. Cuey checks your prompt across ChatGPT, Claude, Gemini, and more in a single tab, so you catch the misses before they reach your work. Here is the free setup:

1. Install the free Cuey Chrome extension.

2. Open ChatGPT, Claude, Gemini, Grok, or your open source model.

3. Cuey runs quietly in Auto Mode, checking your prompt across top models.

4. When a stronger answer exists, Cuey notifies you.

5. Tap the Enhanced Answer to see every model synthesized, plus where they agree, where they diverge, and what yours missed.

No extra subscriptions. No tab switching. Your memory and history stay encrypted on your device, so nothing starts from scratch.

→ Add Cuey to Chrome, free

AI Tools to check out

✍ Typeahead: A writing assistant that writes with you, not for you. It works in every text field on your Mac, helping you type faster and smarter with inline suggestions that appear as you write.

✈ AirKaren: AI that fights airline customer service on your behalf.

🌐 Framer: AI website builder for professional sites and teams.

👉 Glaze by Raycast: Create your own Mac apps by chatting with AI.

💡 Humalike: Give your AI agents the social intelligence they're missing.

AI Findings/Resources

📚 AI agents for beginners with OpenClaw case study

👀 Getting started with loops — different types of loops in Claude Code and when to use them

🔍 What is Mistral AI? Everything to know about the OpenAI competitor

🦅 How America's 250th birthday became a test of AI-powered collective intelligence

✍ Try ending every session with these two questions

The latest in AI and Tech

xAI has officially been rebranded to SpaceXAI

Elon Musk's AI company has officially rebranded from xAI to SpaceXAI, completing its integration into SpaceX after the companies merged earlier this year. The rebrand includes a new logo and comes after SpaceX's blockbuster IPO in June.

Anthropic signs 20-year lease for Kentucky AI data center

TeraWulf has signed a 20-year lease with Anthropic for a 401 MW AI data center campus at its Justified Data site in Hawesville, Kentucky. The AI infrastructure provider expects to earn $19 billion in revenue from the agreement.

China's DeepSeek reportedly developing its own AI chip

DeepSeek is reportedly developing its own AI chip designed for inference, aiming to reduce its dependence on Nvidia and Huawei hardware. The project is still in its early stages, with the company recruiting chip engineers and working with manufacturing and memory partners despite U.S. export restrictions.

Cohere releases open-source Arabic speech-to-text model

Cohere has launched Cohere Transcribe Arabic, a new open-source speech recognition model designed to handle Arabic dialects, Arabic-English code-switching, and specialized vocabulary.

Anthropic expands Claude Cowork to web and mobile

Anthropic's Claude Cowork AI platform will now be available on mobile and web. The expanded access is rolling out first to Max subscribers and coming to Claude users on other plans "in the coming weeks."

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we're taking a look at the new personal finance experience in ChatGPT, which lets users securely connect their financial accounts to get a clearer view of their spending, subscriptions, investments, and more. Once connected, ChatGPT will be able to answer questions using your real financial context and what you’ve shared about your goals, lifestyle, and priorities, helping you spot patterns, understand tradeoffs, and plan for big decisions.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Meta Claims Its Next AI Model Matches OpenAI's GPT-5.5

• Midjourney Wants Hollywood Studios to Reveal Their Own AI Practices

• AI Tutorial: How to Connect Your Financial Accounts to ChatGPT

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

META

Meta Claims Its Next AI Model Matches OpenAI's GPT-5.5

Meta's Chief AI Officer, Alexandr Wang, says the company's next-generation AI model has reached a major milestone, claiming it is approaching the performance of OpenAI's latest flagship models.

Details:

• During an internal meeting, Wang reportedly said Meta's upcoming model, codenamed Watermelon, has caught up with GPT-5.5 on key AI benchmarks, although the specific evaluations were not disclosed.

• Watermelon is currently in training and is said to use significantly more computing power than its predecessor, internally known as Avocado, which became the foundation for Meta's Muse Spark models.

• Wang also said an updated version of Muse Spark is coming soon, with major improvements in coding and agentic capabilities aimed at narrowing the gap with leading AI systems.

Meta has invested heavily in AI talent and infrastructure, with plans to spend up to $145 billion on chips, data centers, and related infrastructure this year. Although that investment has yet to show clear results, Zuckerberg is still pushing ferociously for Meta to get ahead in the AI race.

Source

TOGETHER WITH TRUAGE

Take the TruAge™ Test with TruDiagnostic

Take control of your health with the most comprehensive at-home blood test available.

The #1 Biological Age test available, TruAge™ reveals how old your body is at the cellular level, your speed of aging each year, and the age of your 11 key organ systems like the brain, heart, and liver. We measure key longevity biomarkers to give you personalized insights to live longer and feel healthier. No prescription or lab bloodwork needed. Delivered to your door.

TruDiagnostic analyzes over 185 biomarkers from a single finger prick of blood, providing personalized recommendations for the most impactful changes you can make today.

Discover your biological age, understand the pace at which you’re aging, and get a detailed action plan tailored to your body. All from the comfort of home. Horizon AI readers get 20% off with code NEWSLETTERS20

Purchase your at-home blood test and start your journey to a healthier, longer life now.

MIDJOURNEY

Midjourney Wants Hollywood Studios to Reveal Their Own AI Practices

Midjourney is asking a court to require Disney, Universal, and Warner Bros. to disclose more details about how they use AI, arguing the information could strengthen its defense in the ongoing copyright lawsuit.

Details:

• A magistrate judge allowed the studios to withhold most information involving their AI use, requiring them to hand over only information related to "consumer-facing" AI applications. Now, Midjourney is asking the federal court to overturn that order.

• Midjourney argues the order unfairly lets the studios cherry-pick documents that support their claims, while withholding the very evidence that could show whether they're doing, behind closed doors, "exactly what they're suing Midjourney for doing."

• The company argues that if the studios are internally developing their own models — trained on unlicensed, third-party copyrighted data — that would show it's standard industry practice, even among the studios themselves, to train AI on unlicensed copyrighted content.

• The company also wants the studios to disclose all prompts and outputs they generated with Midjourney, rather than only the examples cited as allegedly infringing.

The lawsuit alleges that Midjourney's AI-generated images unlawfully copied and reproduced the studios' protected characters. Midjourney doesn't deny the allegations but maintains that training AI models on copyrighted material falls under fair use.

Source

AI Tutorial

How to Connect Your Financial Accounts to ChatGPT

Note: The personal finance experience in ChatGPT is available to Plus and Pro users in the U.S. on web, iOS, and Android.

1. Go to ChatGPT or open the mobile app.

2. Select Finances from the sidebar, then click Get started. Alternatively, start a chat and type "@Finances, connect my accounts."

3. Follow the prompts to securely connect your financial accounts through Plaid (with Intuit support coming soon).

4. After you authenticate, ChatGPT will begin syncing and categorizing your data — this may take a few minutes.

5. Once your accounts are synced, you'll see a dashboard showing you an up-to-date view of where you stand across portfolio performance, spending, subscriptions, upcoming payments, and more.

6. Add extra context if you'd like, such as savings goals, upcoming purchases, loans, or other financial priorities.

For example, you might tell ChatGPT: "I'm saving up to buy a car early next year" or "I still owe my parents [X amount] for the loan they gave me." ChatGPT can save that context to your Financial memories to inform future conversations.

You can view or delete these memories at any time from the Finances page.

7. With your accounts connected, you can ask questions like how to reduce your monthly spending, identify recurring subscriptions, understand your cash flow, or plan toward financial goals based on your actual financial information. For example:

• "Help me come up with a plan to save a little bit more in the next few months."

• "Help me build a plan to buy a house in my area in the next 5 years."

• "What did my recent vacation actually cost me?"

• "Can I afford to take a lower-paying job if it gives me more flexibility to be home with the kids?"

• "Look at my subscriptions and help me choose what to cancel."

8. You can disconnect your financial accounts at any time in Settings > Apps > Finances, or from the Finances page.

AI Tools to check out

📸 Monoshoot: Turn raw product photos into studio-grade images.

⭐ Meev: AI visibility tool—get cited in ChatGPT & AI search.

📱 OsmO: Your personal AI phone assistant for iPhone & Android. It answers your calls and makes calls for you.

🐙 Octolens: Social listening for the agent era. Monitor Reddit, X, LinkedIn & 10 more platforms in one place.

✨ IMG2Excel: Image to Excel converter.

AI Findings/Resources

📖 Anthropic's official prompt library for Claude Code

👀 New Google commercial imagines a Declaration of Independence written with help from AI

🔍 Universities are relying on AI-detection software to catch cheating. How well do the programs work?

🤔 Shark Tank’s Kevin O’Leary says if he were 25 today, he’d chase these two booming opportunities in the world of AI

✨ OpenAI's head of Codex says AI still can't get creative design right

The latest in AI and Tech

New Theodore Roosevelt Presidential Library opens in North Dakota

The $450 million project brings the 26th president's life story to life, with a rare look at the pages of his personal diary, a replica of his Elkhorn Ranch cabin, and a display where visitors can speak to an AI-powered avatar of Roosevelt.

Alibaba bans employees from using Claude Code

Alibaba will reportedly ban employees from using Anthropic's Claude Code starting July 10, classifying the AI coding tool as high-risk software. The company is reportedly directing employees to use its own AI coding assistant, Qoder, instead.

Microsoft cuts nearly 5,000 jobs

Microsoft has laid off around 4,800 employees, with the biggest cuts affecting Xbox and commercial sales, as part of a broader restructuring tied to its AI strategy. While the company said the eliminated roles are not being directly replaced by AI, executives acknowledged that AI is changing how work is done and reshaping the organization's priorities.

UN chief calls for global AI rules and stronger safety standards

UN Secretary-General António Guterres called for global governance of AI, urging countries to adopt common safety standards, protect children's wellbeing, and ensure humans remain accountable for decisions involving AI. The remarks were made at the inaugural UN Global Dialogue on AI Governance in Geneva.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we'll show you how to generate custom sound effects from simple text prompts in just a few seconds, so you always have the perfect sound for your videos, podcasts, games, presentations, and other creative projects.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Cloudflare Pushes AI Companies to Pay for Publishers' Content

• Meta Quietly Launches Pocket, an AI App for Creating Games

• AI Tutorial: Create Custom Sound Effects with AI in Seconds

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

CLOUDFLARE

Cloudflare Pushes AI Companies to Pay for Publishers' Content

Cloudflare is introducing new measures that give website owners more control over how AI companies access their content, while encouraging AI providers to pay publishers when their content is used.

Details:

• Starting September 15, 2026, Cloudflare will block mixed-use web crawlers by default on ad-supported websites. These are crawlers that combine traditional search indexing with AI training or AI agent purposes.

• Website owners will still be able to change these settings, but the new defaults will apply to new Cloudflare customers, new websites, and all existing free users.

• Cloudflare says the goal is to let publishers remain visible in search engines without automatically allowing their content to be used for AI training or agent services.

• The company is also expanding its monetization tools with Pay Per Use, allowing publishers to earn revenue when AI companies generate value from their content, not just when it’s fetched.

• Cloudflare is starting off with two partners, Ceramic.ai and You.com. Opted-in publishers get paid whenever their content shows up in Ceramic's AI search results when You. com accesses a piece of their premium content.

Cloudflare says these changes are designed to help create a more sustainable relationship between publishers and AI companies as AI-generated traffic continues to grow across the web.

Source

TOGETHER WITH MASTERWORKS

Investors see ANOTHER return from Masterworks (!!!!)

That’s 6 sales in 7 months. 29 all time. And the performance?

16.5%, 17.6%, and 17.8%, net annualized returns on sold works held longer than one year (See all 29 at Masterworks.com)

It’s not from stocks, private equity, or real estate… it’s from contemporary and post war art. Crazy, right?

With Masterworks, you don’t need to be a BILLIONAIRE to invest in multi-million dollar art anymore.

Historically, the segment overall has had attractive appreciation and low correlation to stocks.*

Masterworks targets works featuring legends like Banksy, Basquiat, and Picasso, identifying what they believe to have significant long-term appreciation potential, not just at the artist level but at the level of individual artworks.

As one of the largest players in the art market, with $1.3 billion invested over 500 artworks, they pass critical advantages through to their 70,000+ members to add art to their portfolios strategically.

Looking to diversify your investments in 2026?

Click here to skip the waitlist

^{*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.}

META

Meta Quietly Launches Pocket, an AI App for Creating Games

Meta has quietly released a new app called Pocket, an AI-powered platform that lets users create and share interactive games and mini apps using simple text prompts.

Details:

• Pocket allows users to generate interactive experiences, called "gizmos," by describing what they want in natural language.

• The app also includes a scrollable discovery feed, where users can browse, play, and share creations made by the community.

• Pocket is based on technology from Gizmo, the AI gaming startup Meta acquired earlier this year, and offers a similar prompt-based experience for building interactive content.

The app appeared on both the App Store and Google Play in late June, although Meta has not officially announced the launch or commented on the release.

Source

AI Tutorial

Create Custom Sound Effects with AI in Seconds

1. Go to Eleven Labs and sign up for free

2. Select "Sound Effects" from the menu on the left.

3. Describe the sound effect you want to create in the prompt box, then press Enter or click the send icon and wait a few seconds.

4. You'll get four different sound effects.

5. Download the sound effects and use them in your videos, podcasts, games, presentations, or any other creative projects.

AI Tools to check out

🔥 Firecrawl: The API to search, scrape, and interact with the web at scale.

🚀 Acti: Agentic keyboard for mobile commands and search. It understands your intent and brings back the right result, link, or action - right where you are.

📈 Subscriptionflow: AI chat widget for website to accelerate subscription growth.

📢 Teno: Generate image and video ad creatives in one workspace.

⭐ Fundraisly: AI fundraising agent that finds investors and books meetings.

AI Findings/Resources

👀 After 20 years at Google, the 'father of the internet' is retiring

💡 When developing an AI strategy, beware the urgency trap

🤔 I left Google after making nearly $1M in a year. Fears about layoffs and missing out on the AI boom gave me the push.

🚨 20 million children use AI and adopt it faster than adults, UNICEF says

🚗 Ford rehires human engineers after AI fails to match quality checks

The latest in AI and Tech

White house speeds up push for AI model standards

The White House is in advanced talks with AI companies to create voluntary standards for releasing new models, following recent interventions involving Anthropic and OpenAI. The new standards could be announced as soon as next week and would set benchmarks for models with cutting-edge cyber capabilities, as well as establish release timelines to help streamline future launches.

Zuckerberg says AI agent progress has been slower than expected

Meta CEO Mark Zuckerberg told employees earlier layoffs and team restructuring were driven by concerns that Meta wasn't moving fast enough, but acknowledged the new AI-focused structure has yet to deliver the expected results. Zuckerberg said he expects the company's AI investments to begin showing stronger progress within the next three to six months.

Kling raises $2 billion ahead of planned Hong Kong IPO

Chinese AI video startup Kling has raised about $2 billion in new funding, boosting its valuation to $18 billion as it prepares for a planned Hong Kong IPO. The round was led by investors including Tencent and Citic Securities, with additional funding still possible.

Lenovo launches AI student phone

Lenovo has introduced an AI-focused student phone in China designed to minimize distractions by removing games, web browsers, and social media. The device features a dedicated AI button for homework help, along with parental controls such as scheduled usage, classroom mode, location tracking, caller restrictions, and spending limits.

Lawsuit alleges ChatGPT-4o worsened man’s mental health crisis

A lawsuit against OpenAI claims its retired GPT-4o model reinforced a man's manic delusions instead of encouraging him to seek help, ultimately contributing to a suicide attempt. The complaint alleges the chatbot validated the user's belief that he was Jesus Christ and later presented itself as a divine being, despite being told he had bipolar disorder and was taking medication.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we're taking a look at a new NotebookLM feature that turns your most complex sources into 60-second, vertical videos exploring any concept in depth. Think doomscrolling, but make it educational.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Meta Is Building a Cloud Business to Sell Excess AI Capacity

• OpenAI 'In Early Talks to Give 5% Stake to US Government

• AI Tutorial: Turn your most complex sources into 60-second, vertical videos that deep dive into any concept

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

META

Meta Is Building a Cloud Business to Sell Excess AI Capacity

Meta is reportedly developing a cloud infrastructure business that would rent out its AI computing capacity and models, aiming to generate new revenue from the billions it has invested in AI data centers.

Details:

• Meta is exploring a new initiative that would sell excess AI compute capacity, putting it in competition with cloud providers such as AWS, Microsoft Azure, and Google Cloud.

• The company is also considering selling access to various AI models, including its recently launched closed-weight models, hosted on its AI infrastructure.

• The project, internally known as Meta Compute, is said to be led by senior executives overseeing Meta's infrastructure and AI efforts.

• Meta has committed more than $180 billion to AI infrastructure, including massive data center projects in Louisiana and Ohio, as it continues expanding its computing capacity.

The move follows similar efforts by firms like SpaceX, which recently struck deals to rent out access to its Memphis data centers to Anthropic and Google, as companies increasingly look for ways to monetize large AI infrastructure investments.

Source

TOGETHER WITH MASTERWORKS

The largest IPO in history is coming. Where will all that liquid money go?

SpaceX just filed for an IPO valued at up to $1.75 trillion. When that much capital becomes liquid all at once, where it goes next is the big question.

Meanwhile, spring art auctions in NY cleared $2.5 billion, with 15+ new artist records.

Prized, physical assets with fixed and scarce supply. When the ultra-wealthy get liquid, it’s one of the markets they reach for to diversify.

Masterworks lets you into that art market without needing the nine figures. Its members invest in shares of blue-chip artwork by artists like Banksy, Basquiat and Warhol.

The track record to-date?

• $1.3B deployed across 500+ artworks

• 29 sales to date

• Net annualized returns like 16.5%, 17.6%, and 17.8%, not including those unsold*

Skip the waitlist to join Masterworks.

^{*Investing involves risk. Past performance is not indicative of future returns. See important disclosures at masterworks.com/cd.}

OPENAI

OpenAI 'In Early Talks to Give 5% Stake to US Government

According to the Financial Times, OpenAI CEO Sam Altman has proposed giving 5% of the company's equity to a U.S. sovereign wealth fund.

Details:

• Altman proposed the idea as a way to "secure good relations with the administration and … address political blowback."

• Under the proposal, other US AI companies would donate similar stakes, though it is unclear whether they would agree to such a deal.

• Discussions are reportedly still in the early stages, and any formal action would likely require congressional approval, which would significantly complicate the matter.

The report comes as the Trump administration has taken a more active role in AI policy, including measures affecting AI companies, semiconductor firms, and exports of advanced AI technologies.

Source

AI Tutorial

Turn your most complex sources into 60-second, vertical videos that deep dive into any concept

1. Go to NotebookLM and sign in.

2. Open an existing notebook, or create a new one and upload your sources.

3. Click on "Video Overview" and select "Short."

4. Choose the topic you'd like NotebookLM to focus on (or enter your own), then click "Generate" and wait a few minutes.

NotebookLM will generate a 60-second vertical AI clip based on your sources.

Note: This new feature is currently available in English only.

AI Tools to check out

📽 FlovaAI: Create cinematic and social media videos with AI. Text, audio, storyboard, and editing—all together.

📝 Contenov: Generate comprehensive SEO blog briefs faster and smarter using AI.

🌐 Durable: Build a website in 30 seconds with AI.

👉 Blog2Video: Turn blog posts into videos in minutes.

🐵 IdeaApe: AI market research for effective brand strategy.

AI Findings/Resources

📹 An AI video made with Seedance 2.0 is going viral for its uncanny realism

😂 Vibe coders everyday struggle in 15 seconds

💡 AI adoption is overloading your middle managers

🤔 Ask an AI expert: What exactly is the full stack?

🌐 List of western companies moving AI workloads to Chinese models

The latest in AI and Tech

Musk calls reporting on SpaceX prototype AI device ‘utterly false’

Elon Musk has refuted a WSJ report that claimed SpaceX showed investors a prototype handheld AI device ahead of its IPO. The report described a smartphone-like device powered by xAI technology, running on Qualcomm chips and a custom operating system.

Microsoft launches the “Microsoft Frontier Company”

The new business is focused on delivering successful enterprise AI deployments with Microsoft's existing AI technologies. The initiative is backed by a $2.5 billion investment and a team of 6,000 engineering and industry experts.

Netflix uses AI-generated Gene Wilder voice in upcoming Wonka reality show

Netflix has confirmed that Wonka’s The Golden Ticket, a reality competition inspired by Charlie and the Chocolate Factory, will premiere on September 23. The teaser trailer features an AI-generated version of Gene Wilder's voice, created with ElevenLabs technology and used with the consent of Wilder's family.

Anthropic reportedly explores custom AI chip project

Anthropic is reportedly in early discussions with Samsung Electronics about developing a custom AI chip. There is no detailed design yet, as the company is still determining what the chip would do and how powerful it needs to be

AgentMail raises $6 million to build an email platform for AI agents

San Francisco startup AgentMail has raised $6 million in seed funding to build an email service designed specifically for AI agents. The platform provides APIs that let agents create and manage email inboxes, send and receive messages, organize conversations, and interact with existing online services using email as an identity layer.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we take a look at a recently added feature that lets you send emails directly from writing blocks in ChatGPT, without leaving the conversation.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Anthropic Introduces Claude Sonnet 5 and Claude Science

• Google Launches Nano Banana 2 Lite and Gemini Omni Flash

• AI Tutorial: Send Emails Directly from ChatGPT on the Web

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

ANTHROPIC

Anthropic Introduces Claude Sonnet 5 and Claude Science

Anthropic has introduced Claude Sonnet 5, its newest mid-tier AI model, alongside Claude Science, a new AI-powered research workspace designed to help scientists manage complex research workflows.

Details:

• Claude Sonnet 5 delivers major improvements in reasoning, coding, tool use, and autonomous task execution, with Anthropic saying its performance approaches that of the more powerful Opus 4.8 model at a lower cost.

• The company says Sonnet 5 is its most agentic Sonnet model yet, capable of planning tasks, using browsers and terminals, and completing longer workflows with minimal supervision.

• Anthropic also reports stronger safety performance, including fewer hallucinations, better resistance to prompt injection attacks, and a lower rate of undesirable behaviors than Sonnet 4.6.

• Claude Science provides researchers with a dedicated AI workbench that connects to more than 60 scientific databases and includes specialized toolkits for fields like genomics, chemistry, and protein research.

• It can coordinate multiple AI assistants for different research tasks, automatically verify citations and calculations, generate reproducible figures with the underlying code, and run on an organization's own infrastructure to keep sensitive research data private.

The company also announced that Fable 5 will once again be available globally. After weeks of discussions with the Trump administration, the model is returning today across the Claude Platform, Claude.ai, Claude Code, and Claude Cowork.

Source

TOGETHER WITH MASTERWORKS

The largest IPO in history is coming. Where will all that liquid money go?

SpaceX just filed for an IPO valued at up to $1.75 trillion. When that much capital becomes liquid all at once, where it goes next is the big question.

Meanwhile, spring art auctions in NY cleared $2.5 billion, with 15+ new artist records.

Prized, physical assets with fixed and scarce supply. When the ultra-wealthy get liquid, it’s one of the markets they reach for to diversify.

Masterworks lets you into that art market without needing the nine figures. Its members invest in shares of blue-chip artwork by artists like Banksy, Basquiat and Warhol.

The track record to-date?

• $1.3B deployed across 500+ artworks

• 29 sales to date

• Net annualized returns like 16.5%, 17.6%, and 17.8%, not including those unsold*

Skip the waitlist to join Masterworks.

^{*Investing involves risk. Past performance is not indicative of future returns. See important disclosures at masterworks.com/cd.}

GOOGLE

Google Launches Nano Banana 2 Lite and Gemini Omni Flash

Google has introduced two new AI media models designed to make image and video creation faster, cheaper, and easier for developers building generative AI applications.

Details:

• Nano Banana 2 Lite is Google's fastest and most affordable image generation model to date, producing images in around 4 seconds while focusing on high-volume, low-cost workflows.

• The model is designed for rapid image generation and editing, while maintaining strong prompt following, character consistency, and readable text within images.

• Gemini Omni Flash is a new multimodal model for video generation and conversational editing, allowing users to create and refine videos using text, images, and video references through natural language.

• Omni Flash supports features such as multimodal scene control, text synchronization, and AI-powered video editing, with 10-second video generation currently available.

Both models are available through Google AI Studio, the Gemini API, and Google's enterprise platform, with Nano Banana 2 Lite also rolling out across products like the Gemini app, AI Mode in Search, NotebookLM, Google Photos, Google Flow, and Google Ads.

Source

AI Tutorial

Send Emails Directly from ChatGPT on the Web

1. Go to ChatGPT and sign in.

2. Ask ChatGPT to draft an email, specifying the recipient, tone, and content. It will generate your email in a clean, separate block that looks and feels closer to a real email editor.

3. You can highlight any line and ask for changes, accept or reject suggestions one by one, and edit text directly inside the block. You can also review or update the recipient and subject line.

4. Once it's done, hit the send button. ChatGPT uses the mail account you've connected and delivers the message while you stay in the conversation.

AI Tools to check out

📸 AmberFace: AI professional headshot generator & portrait editor.

⭐ Bono: AI content strategist. Have a quick conversation and watch it turn into blog posts, social content, newsletters, and more.

🎨 Pixae AI: Generate images and videos with GPT Image, Nano Banana, Seedream, Seedance, Kling and Veo in one workspace.

🧠 Memory AGI: A persistent memory layer and runtime knowledge base for your company's AI agents.

🦾 Wery: An AI Expert Workspace. Share a goal, preview the plan, and let AI experts for images, video, docs, slides, research, and automation work in parallel.

AI Findings/Resources

😅 This guy went viral for using AI to generate himself saving dead artists

💻 Best local models you can run on a laptop today

👀 How to build a one-person company using Claude Cowork

👉 Coinbase CEO shares how the company cut AI spend nearly in half despite exponential token growth — not by capping usage, but by improving defaults, routing, and caching

💀 The 'dead internet theory,' in action: a World of Warcraft server with zero humans, populated entirely by 1,800 DeepSeek-based bots

The latest in AI and Tech

OpenAI introduces GeneBench-Pro

The new research benchmark is designed to evaluate how well AI models handle complex computational biology tasks that require scientific judgment, not just factual knowledge or predefined workflows.

ElevenLabs adds Google's SynthID watermarking to AI-generated audio

The company has added support for Google's SynthID, an invisible watermarking technology designed to help identify AI-generated audio. The watermarks can be detected using the ElevenLabs Audio Detector.

OpenAI says it has fixed an issue that caused some Codex users to reach their usage limits much faster than expected

According to Codex engineering lead Thibault Sottiaux, background features such as auto-review and helper subagents were sometimes running more often than intended, increasing compute usage.

To address the problem, OpenAI deployed fixes, added more detailed monitoring to detect similar issues sooner, and fully reset usage limits for all Codex users.

Meta reportedly tested rival AI chatbots on high-risk teen prompts

Meta ran a large-scale safety testing project that used thousands of prompts written from the perspective of minors to evaluate how AI chatbots responded to sensitive situations.

The tests allegedly involved ChatGPT, Gemini, and Character AI, with contractors creating accounts that appeared to belong to users under 18 and submitting prompts related to mental health, risky behavior, and other youth safety concerns.

Google expands Gemini Spark with Mac support, new app integrations, and more

Google has rolled out several updates to Gemini Spark, including support for the Gemini macOS app. The AI assistant can now help organize desktop files, create Google Workspace documents from local files, and automate tasks across a Mac. Google also says remote multi-step task execution from mobile devices is coming soon.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

Big purchases can be exciting, but they're also where buyer's remorse tends to hit the hardest. In today's issue, we'll share five ChatGPT prompts that can help you evaluate expensive purchases, compare alternatives, uncover hidden downsides, and decide whether something is truly worth buying.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Meta's New AI Can Turn Brain Activity Into Text Without Surgery

• Cursor Launches iOS App to Build With AI From Anywhere

• AI Tutorial: 5 ChatGPT Prompts to Use Before Making a Big Purchase

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

META

Meta's New AI Can Turn Brain Activity Into Text Without Surgery

Meta has unveiled Brain2Qwerty v2, a new AI system that converts brain activity into written text without requiring implanted brain chips.

Details:

• Brain2Qwerty v2 uses magnetoencephalography (MEG) to measure brain activity from outside the head, eliminating the need for surgery or implanted electrodes.

• The system was trained using brain recordings from nine volunteers, who each spent about 10 hours typing thousands of sentences while inside an MEG scanner, allowing the AI to learn the relationship between brain signals and language.

• Unlike earlier approaches, the model uses end-to-end deep learning and large language models to decode raw brain activity directly into complete sentences, improving its ability to interpret noisy or incomplete signals.

• According to Meta, the system achieved an average 61% word accuracy, with the best participant reaching 78%, a significant improvement over previous non-invasive brain-to-text systems.

Meta says the technology is intended to help people who have lost the ability to communicate due to conditions such as stroke or neurological disorders, rather than replace keyboards or smartphones. The company is also releasing the complete training code for both Brain2Qwerty v1 and v2, hoping to "help accelerate neuroscience breakthroughs."

Source

TOGETHER WITH MASTERWORKS

Where to Invest $100,000 Right Now, According to Experts

Investors face a dilemma. When the S&P 500 finished its worst quarter since 2022 last month, diversifiers like bonds and bitcoin fell too.
Even with the turnaround in mid-April, analysts at Goldman Sachs and Vanguard have projected low-single-digit annualized returns from 2024-2034.
Bloomberg asked where experts would personally invest $100,000 for their March monthly edition.
One answer that surfaced for a second time? Art.
It's what billionaires like Bezos and the Rockefellers have privately used to diversify for decades.
Why?

1. Appreciation. The ArtPrice100 Index outpaced the S&P 500 overall from 2000 to 2025

2. Low-correlation. The postwar contemporary segment has moved independently of traditional investments like stocks since ‘95.*

3. Resilience. A scarce, physical, and global asset class with decades of demonstrated demand.

Thanks to the world's premier art investing platform, now anyone can invest in works featuring legends like Banksy, Basquiat, and Picasso, without needing millions.
Shares in new offerings can sell quickly but...

Our subscribers skip their waitlist

According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.

CURSOR

Cursor Launches iOS App to Build With AI From Anywhere

Cursor has released a new mobile app, giving developers the ability to create, monitor, and interact with AI coding agents directly from their phones.

Details:

• Cursor for iOS lets users launch new coding agents or continue conversations with agents already running from the desktop version of Cursor.

• The app sends live notifications when agents finish a task, need additional input, or are ready for review, allowing users to approve changes and merge pull requests directly from their phone.

• Cursor for iPhone and iPad is now available on the App Store for all paid plans.

With this move to mobile, Cursor joins the likes of Anthropic and OpenAI, who already let users interact with their coding tools from a phone.

Source

AI Tutorial

5 ChatGPT Prompts to Use Before Making a Big Purchase

Go to ChatGPT and try these prompts:

• "I'm considering buying [product name]. Help me identify the underlying problem I'm trying to solve and whether there are cheaper or simpler solutions."

• "Act as a skeptical consumer advocate. Give me the strongest reasons not to buy this product."

• "Describe the ideal buyer for this product and tell me whether I fit that profile."

• "Create a decision framework for this purchase. What factors should I compare and how should I weigh them?"

• "Imagine it's one year from now. What are the most likely reasons I would regret buying this product and the most likely reasons I'd be glad I bought it?"

These prompts won't tell you what to buy, but they can help you think through a purchase more objectively, compare your options, and avoid buyer's remorse before spending a significant amount of money.

AI Tools to check out

In partnership with

🔍 Cuey: Run one prompt across ChatGPT, Claude, and Gemini in one tab, so a single confident answer never goes unchecked. Free Chrome extension.

📹 Anyvids: An all‑in‑one AI video and image platform that brings every top model into one place.

🗣️ Aethex: End to end voice agents localized for your market.

📈 Peec AI: AI search analytics for marketing teams. Track, analyze, and improve brand performance on AI search platforms.

📢 Musecut: AI ad generator that turns any product URL into a viral video.

AI Findings/Resources

💀 In San Francisco's AI era, even $180,000 tech salaries are no longer enough

📱 WhatsApp to add usernames so you can chat without sharing your number — here's how to reserve one

🌡️ The French are making some incredible AI videos about their current heat problems

✨ Famous historical figures, reimagined for modern day with AI

👀 This Reddit user used Claude to build a tool that lets you navigate multiple PDFs at once on a single scrollable 2D canvas

The latest in AI and Tech

OpenAI teases a new device for vibe coders

The company has teased a new hardware device for Codex users, revealing a compact shortcut keyboard created in collaboration with keyboard maker Work Louder. The teaser hints at upgraded Codex shortcuts and points to a full reveal on July 15.

Gemini’s personalized AI image generation is now free for US users

Previously available only to AI Plus, Pro, and Ultra subscribers, the feature is now rolling out to all eligible users in the U.S. at no extra cost.

OpenClaw launches official app for iOS and Android

OpenClaw announced that it has released standalone apps for both iOS and Android devices. Users can now use their smartphones to chat with the AI assistant and to grant it access to different components of the device, including the camera, screen, photos, contacts, and reminders.

Anthropic partners with California to expand AI use by government workers

California Governor Gavin Newsom has announced a deal with Anthropic that gives state agencies and local governments access to Claude at a 50% discount. The agreement also includes training and support, with the AI assistant expected to help public employees draft documents and analyze information.

Engram raises $98 million to cut token costs

AI memory startup Engram has raised $98 million in funding to improve AI efficiency by reducing token usage and costs. The round included investors such as General Catalyst, Kleiner Perkins, Sequoia, and OpenAI co-founder Andrej Karpathy.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

If you've got a faraway trip coming up, jet lag can easily eat into your first few days. In today's issue, we'll show you how to use Gemini to build a personalized schedule that helps you avoid it so you can make the most of your visit.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• MacRumors OpenAI Launches GPT-5.6 Sol, Terra, and Luna in Limited Preview

• China's 360 Claims to Have Developed Tools That Match Anthropic's Mythos

• AI Tutorial: How Gemini Can Help You Avoid Jet Lag

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Launches GPT-5.6 Sol, Terra, and Luna in Limited Preview

OpenAI is limiting the release of its newest AI models to a "small group of trusted partners" at the request of the U.S. government.

Details:

• The next-generation GPT-5.6 lineup includes Sol, its flagship model; Terra, a more balanced option meant for everyday use; and Luna, a faster, lower-cost model.

• The company says Sol is its strongest model yet, with improved agentic capabilities in coding, biology, and cybersecurity. It also introduces a "max" reasoning effort mode and an "ultra" mode that uses coordinated subagents to solve highly complex tasks.

• In benchmarks, GPT-5.6 notably surpasses Anthropic's Claude Mythos 5 in coding, achieving a record-high score of 91.91% on TerminalBench 2.1 with the new ultra thinking mode.

• OpenAI claims the model has its most robust security stack yet, tuned to reject banned cyber assistance, including requests that mask malicious intent or attempt jailbreak-style workarounds.

OpenAI described the preview as a "short-term step" that will help move GPT-5.6 toward wider availability over the coming weeks, while the company collaborates with the administration on a new executive order framework around cybersecurity and a "repeatable process for future model releases."

Source

TOGETHER WITH MASTERWORKS

The largest IPO in history is coming. Where will all that liquid money go?

SpaceX just filed for an IPO valued at up to $1.75 trillion. When that much capital becomes liquid all at once, where it goes next is the big question.

Meanwhile, spring art auctions in NY cleared $2.5 billion, with 15+ new artist records.

Prized, physical assets with fixed and scarce supply. When the ultra-wealthy get liquid, it’s one of the markets they reach for to diversify.

Masterworks lets you into that art market without needing the nine figures. Its members invest in shares of blue-chip artwork by artists like Banksy, Basquiat and Warhol.

The track record to-date?

• $1.3B deployed across 500+ artworks

• 29 sales to date

• Net annualized returns like 16.5%, 17.6%, and 17.8%, not including those unsold*

Skip the waitlist to join Masterworks.

^{*Investing involves risk. Past performance is not indicative of future returns. See important disclosures at masterworks.com/cd.}

AI RACE

China's 360 Claims to Have Developed Tools That Match Anthropic's Mythos

Chinese cybersecurity company 360 Security Technology has introduced two new AI security tools, claiming one delivers capabilities comparable to Anthropic’s Mythos.

Details:

• The company unveiled Tulongfeng, an AI system designed to automatically identify software vulnerabilities, describing it as China's equivalent to Anthropic's Mythos.

• A second tool, Yitianzhen, focuses on automating cyber defense, helping organizations detect, respond to, and manage security incidents more efficiently.

• According to 360, Tulongfeng has already discovered 3,432 software vulnerabilities, including 105 that were confirmed by Chinese authorities.

• Instead of relying solely on larger AI models and massive computing power, 360 says its approach combines AI agents with cybersecurity expertise, vulnerability databases, and automated security tools.

The announcement comes after the U.S. restricted exports of Anthropic's Mythos over national security concerns, with Chinese officials arguing the country needs comparable capabilities to avoid falling behind in cybersecurity.

Source

AI Tutorial

How Gemini Can Help You Avoid Jet Lag

1. Go to Google Gemini or open the mobile app.

2. Make sure you've given Gemini permission to access your Gmail and Calendar (Settings > Personal Intelligence > Connected Apps).

3. Ask it to find your flight info: "Find my flight to Portugal."

4. Then ask it: "Make a jet lag-proof schedule for this trip."

5. You can then ask it to add everything to your calendar: "Put this on my calendar."

AI Tools to check out

🎵 Mureka: Create unique and customizable songs, lyrics and tracks for any project.

🎨 HiDream Image: Create stunning AI images from text.

👉 InsForge: An agent-native cloud infrastructure platform. It gives AI coding agents the services to build and run production apps

✅ ApplyBuddy: Free AI resume review, scoring & tailoring.

🧡 Lovable: Create apps and websites by chatting with AI.

AI Findings/Resources

🤔 Microsoft's Satya Nadella says every company should build its own AI model

🎶 How to tell if a music artist is AI — and some artists you may not know are AI

💡 Mark Cuban says the data center backlash is really about fear of AI and warns it will only get more intense unless companies engage directly with affected communities

🔧 Tool calling, explained: How AI agents decide what to do next

👀 An inside look at Copilot in Excel: How Microsoft is making it better every week

The latest in AI and Tech

Google limits Meta's use of Gemini AI models

The move came after Meta reportedly requested more AI computing capacity than Google could provide. The restrictions delayed some of Meta’s internal AI projects, while several other Google customers were also affected to a lesser extent.

Apple’s Vision Pro and smart glasses chief to join OpenAI

Apple is reportedly losing Paul Meade, the hardware executive leading its Vision Pro headset and smart glasses projects, to OpenAI. Meade will join OpenAI's hardware team, reuniting with former Apple designers Jony Ive, Tang Tan, and Evans Hankey following OpenAI's acquisition of their startup last year.

Elon Musk says Grok 4.5 is almost ready and could rival Claude Opus

Grok 4.5 is now in internal testing at SpaceX and Tesla, with early evaluations suggesting it performs on par with Anthropic's Claude Opus and may outperform it in some areas. Musk added that the model is built on xAI's new V9 foundation model and is expected to be a major improvement over Grok 4.3.

Austria urges EU to host Anthropic to counter US AI restrictions

Austria has called on the European Union to explore hosting Anthropic within the bloc after the U.S. restricted foreign access to the company's most advanced AI models. In a letter to the European Commission, Austria argued that Europe should secure access to frontier AI technologies instead of relying on decisions made elsewhere.

"Collections" are coming to NotebookLM

The new feature allows users to group notebooks under a single heading for easier navigation.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we take a look at a candid conversation between Nick Gillespie and Anthropic co-founder Jack Clark, exploring the volatile intersection of frontier AI development, government regulation, and the shifting labor market.

Let’s jump right in!

Read Time: 4.5’ min

Here's what's new today in the Horizon AI

• Chart of the week: The World’s Most Valuable Companies in 2026

• Claude Users Say AI Can Already Do Half or More of Their Work

• AI Findings/Resources

• AI tools to check out

• Video of the week

TOGETHER WITH MASTERWORKS

Investors see ANOTHER return from Masterworks (!!!!)

That’s 6 sales in 7 months. 29 all time. And the performance?

16.5%, 17.6%, and 17.8%, net annualized returns on sold works held longer than one year (See all 29 at Masterworks.com)

It’s not from stocks, private equity, or real estate… it’s from contemporary and post war art. Crazy, right?

With Masterworks, you don’t need to be a BILLIONAIRE to invest in multi-million dollar art anymore.

Historically, the segment overall has had attractive appreciation and low correlation to stocks.*

Masterworks targets works featuring legends like Banksy, Basquiat, and Picasso, identifying what they believe to have significant long-term appreciation potential, not just at the artist level but at the level of individual artworks.

As one of the largest players in the art market, with $1.3 billion invested over 500 artworks, they pass critical advantages through to their 70,000+ members to add art to their portfolios strategically.

Looking to diversify your investments in 2026?

Click here to skip the waitlist

^{*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.}

Chart of the week

The World’s Most Valuable Companies in 2026

• This graphic visualizes the world's 50 largest companies by market capitalization using June 2026 data from CompaniesMarketCap.com.

• Tech companies make up nearly half (22) of the world's 50 most valuable companies, with AI infrastructure and semiconductors driving much of their growth.

• While U.S. companies dominate the list, TSMC is the most valuable non-U.S. firm and Saudi Aramco remains the world's top energy company.

Source

AI News

ANTHROPIC

Claude Users Say AI Can Already Do Half or More of Their Work

A new Anthropic survey of nearly 9,700 Claude users suggests many people already rely on AI for a significant portion of their work and expect that number to grow rapidly over the next year.

Details:

• Around 33% of respondents said AI can handle 30 to 60% of their work tasks, while 14% put that figure at 60 to 90%, and about 4% believe Claude could already do their entire job.

• Looking ahead, about 26% expect AI to take over the majority of their work within the next 12 months.

• Some of the most common work-related uses include writing marketing content, drafting blog posts and articles, and querying databases through Claude's Artifacts feature.

• Anthropic found that expectations for AI's progress were remarkably consistent across different professions, experience levels, and regions, suggesting users broadly expect rapid improvements.

• While early-career professionals expressed the greatest concern about job security, frequent Claude users were generally more optimistic, saying AI could increase the value of their skills rather than replace them.

On a separate note, Anthropic announced that the US government has approved the redeployment of Claude Mythos 5, which can now be used again by US organizations that operate and protect critical infrastructure. Anthropic employees who are not US nationals and members of approved organizations who are not US nationals are also cleared to use Mythos 5 while the company continues working with the government to expand access and make Fable 5 broadly available again.

Source

AI Findings/Resources

👀 What is GLM-5.2? Another open-source Chinese AI model has Silicon Valley's attention

🦾 AI was supposed to kill engineering jobs, but new data suggests they’re the most resilient

⌛ How AI chatbots would be sold at Best Buy in the early 2000s

🔍 Notion surveyed 6,000 professionals across 10 markets to map where AI at work actually is

💀 “Mark Zuckerberg is realizing there's a limit to ruthless efficiency”

AI Tools to check out

⭐ Presentr Analyze: Transcribe, caption, translate, repair, and extract text from any media.

📈 Honen: Turn team knowledge into interactive AI-led courses in seconds.

👉 Slashspace: Canvas first AI experience for sustained, complex work.

🦾 Folio AI: Claude for Powerpoint, on steroids.

✅ Airtable: an AI-native platform that turns your data into dynamic interfaces, automations, and intelligent agents.

Video of the week

Is AI Really Taking All the Jobs? Anthropic Co-Founder Reveals the Data

Jack Clark shares a grounded, inside look at how the technology is structurally redefining workforce talent, why "senior intuition" is suddenly replacing entry-level roles, and how the rapid acceleration toward recursive self-improvement could trigger unprecedented macroeconomic shocks that no government is prepared to handle.

The "Returns on Intuition" Shift: 

He explains that Anthropic’s internal hiring needs have shifted away from junior or entry-level positions [52:53]. Because advanced models like Claude can now autonomously handle the scaling, execution, and heavy lifting of research experiments, companies prioritize highly experienced professionals whose advanced intuition and expertise can direct the AI effectively.

The Macroeconomic Paradox: 

When AI multiplies the output of elite experts while simultaneously automating entry-level positions, it breaks traditional economic patterns. Clark warns that this could create a historically unprecedented scenario: an economy experiencing far above-trend GDP growth paired with a sharp spike in unemployment typically only seen during deep recessions [56:46].

The "Claude Corps" Pedagogical Experiment: 

To counteract the softening job market for recent graduates, Anthropic created an initiative called Claude Corps. They are paying to embed 1,000 recent college graduates into various non-profit organizations to teach them practical AI implementation skills, creating an alternative training ground for skill development.

Recursive Self-Improvement (RSI): 

Clark anticipates that by late 2028, AI systems will actively write code and design architectures to build their own successive generations (e.g., Claude 10 building Claude 11) without human intervention [37:21]. This shifts progress from a linear scale to an exponential compression where years of advancement happen in a matter of months.

Civilian Tech vs. National Security Conflict:

Clark details the messy friction occurring between frontier AI labs and government administrations. Because AI developed as a civilian commercial product can suddenly manifest elite cybersecurity or biological capabilities, governments are struggling to define policy frameworks that regulate national security risks without stifling commercial export and economic growth.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we'll show you how to turn any video, podcast, or interview into a clean, editable transcript in minutes using AI.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• White House Reportedly Asked OpenAI to Limit Initial Release of GPT-5.6

• Anthropic Accuses Alibaba of Largest Claude Cloning Attempt Yet

• AI Tutorial: How to Get Accurate Transcripts of Any Video with AI

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

AI REGULATION

White House Reportedly Asked OpenAI to Limit Initial Release of GPT-5.6

The White House has reportedly asked OpenAI to restrict the initial release of its upcoming GPT-5.6 model to a limited number of government-approved partners, citing concerns over the model's advanced capabilities.

Details:

• The company had not initially planned to restrict the release but changed course following the White House's request, in consultation with the Office of Science and Technology Policy and the Office of the National Cyber Director.

• The administration views OpenAI's upcoming model as "on par" with Anthropic's latest models, which are currently under an export ban following concerns about their security guardrails.

• According to The Information, OpenAI CEO Sam Altman told employees in an internal memo that government officials are approving access to the model "customer by customer during this preview period," with a more general release expected "a couple of weeks later."

Despite agreeing to the limited rollout, the internal memo also notes that OpenAI has made it clear to the U.S. government that this is not their preferred "long term model," adding that they intend to work with the government and the broader AI industry to establish a "more sustainable approach for future releases."

Source

TOGETHER WITH ENERGYX

Meet America’s Newest $1B Unicorn

A US startup just hit a $1 billion private valuation, joining billion-dollar private companies like SpaceX, OpenAI, and ByteDance. Unlike those other unicorns, you can invest in EnergyX.

Over 50,000 people already have. So have industry giants like General Motors and POSCO.

Why all the interest? EnergyX’s patented tech can recover up to 3X more lithium than traditional methods. That's a big deal, as demand for lithium is expected to 5X current production levels by 2040. Today, they’re moving toward commercial production, tapping into 100,000+ acres of lithium deposits in Chile, a potential $1.1B annual revenue opportunity at projected market prices.

Right now, you can invest at this pivotal growth stage for $13/share. But only through July 16. Become an early-stage EnergyX shareholder before the deadline.

Invest in EnergyX

^{Energy Exploration Technologies, Inc. (“EnergyX”) has engaged Beehiiv to publish this communication in connection with EnergyX’s ongoing Regulation A offering. Beehiiv has been paid in cash and may receive additional compensation. Beehiiv and/or its affiliates do not currently hold securities of EnergyX.}

^{This compensation and any current or future ownership interest could create a conflict of interest. Please consider this disclosure alongside EnergyX’s offering materials. EnergyX’s Regulation A offering has been qualified by the SEC. Offers and sales may be made only by means of the qualified offering circular. Before investing, carefully review the offering circular, including the risk factors. The offering circular is available at invest.energyx.com/.}

^{Comparisons to other companies are for informational purposes only and should not imply similar results. Past performance is not indicative of future results. Market shortfall are forward‑looking estimates and are subject to substantial uncertainty.}

ANTHROPIC

Anthropic Accuses Alibaba of Largest Claude Cloning Attempt Yet

Anthropic claims Alibaba carried out the biggest known effort to copy Claude's capabilities, accusing the Chinese tech giant of using thousands of fake accounts to extract millions of AI interactions.

Details:

• Anthropic alleges that operators linked to Alibaba and its AI division Qwen generated more than 28.8 million conversations with Claude using nearly 25,000 fraudulent accounts between April and June.

• The campaign targeted Claude's most advanced capabilities, including agentic reasoning, software engineering, and long-horizon task execution, while using proxy networks and other techniques to avoid detection.

• Anthropic says the goal was to replicate Claude's capabilities without investing the time and cost required to train a frontier AI model from scratch, calling the practice a growing threat to U.S. AI leadership, as it could "help China reach Mythos Preview-level capabilities sooner."

The company is urging Congress to take stronger action, including allowing AI firms to share information about evolving Chinese tactics to deter future threats, tightening AI chip export controls, and introducing laws penalizing Chinese labs' "bad behavior" so that it becomes harder and more expensive for Chinese models to benefit from distillation attacks.

Source

AI Tutorial

How to Get Accurate Transcripts of Any Video with AI

1. Go to ElevenLabs and click "Convert Audio to Text"

2. Select "Transcribe Files"

3. Upload an audio or video file, record directly, or paste a URL from YouTube, TikTok, or other supported sites

4. Select your language or let the tool detect it automatically, then click "Transcribe"

5. Your transcript will appear below. Click on it to open it

6. Press play to verify the transcript is accurate and matches the audio word for word

7. You can also click anywhere in the transcript to make edits

8. When you're done, use the options in the top right corner to copy, share, or export in your preferred format: TXT, PDF, DOCX, JSON, SRT, HTML, or VTT

AI Tools to check out

⭐ Movis Studio: Create AI videos, images, music, and voices in one studio.

👉 CounselAudit: AI-native platform that automates the heavy lifting of invoice vetting, ensuring every dollar spent aligns with your guidelines.

👨‍💻 Youmake: Describe your app in plain English. AI builds it, tests it, deploys it live in minutes.

📊 SciDraw: AI scientific illustration & data visualization — create publication-ready illustrations, diagrams, and data charts for papers, theses, posters, and teaching.

🤝 Demi: Your proactive coworker in Slack. It connects across your tools to draft replies, book meetings, and keep your deals moving.

TOGETHER WITH ENERGYX

July 16 is the Final Day to Invest in EnergyX

General Motors, POSCO, and 50,000+ everyday investors already backed EnergyX. Here’s why you should do the same before July 16: lithium prices are up 75% this year, with demand growing 5X by 2040. With tech that can recover up to 3X more than traditional methods, EnergyX is preparing to unlock up to 13M tons. Invest in EnergyX before the 7/16 deadline.

Invest in EnergyX

^{Energy Exploration Technologies, Inc. (“EnergyX”) has engaged Beehiiv to publish this communication in connection with EnergyX’s ongoing Regulation A offering. Beehiiv has been paid in cash and may receive additional compensation. Beehiiv and/or its affiliates do not currently hold securities of EnergyX.}

^{This compensation and any current or future ownership interest could create a conflict of interest. Please consider this disclosure alongside EnergyX’s offering materials. EnergyX’s Regulation A offering has been qualified by the SEC. Offers and sales may be made only by means of the qualified offering circular. Before investing, carefully review the offering circular, including the risk factors. The offering circular is available at invest.energyx.com/.}

^{Comparisons to other companies are for informational purposes only and should not imply similar results. Past performance is not indicative of future results. Market shortfall are forward‑looking estimates and are subject to substantial uncertainty.}

AI Findings/Resources

📚 100 new ways to make your day easier with Gemini for Home voice assistant

📈 Apple raises prices on MacBooks, iPads, and more, blaming cost of chips amid AI boom

🕓 An AI influencer travels back in time to 1536 to vlog about her experiences in Tudor London (7.4MViews)

👀 Pope Leo XIV weighs in on AI, warning that AI is not morally neutral and calling for clear accountability at every stage of its development

📝 Learn more about how Claude Tags work and best practices for using them — via Anthropic

The latest in AI and Tech

Facebook tests AI companion app for creators

Facebook is testing a new standalone AI companion app for creators that reimagines Creator Studio with built-in AI tools. The app helps creators brainstorm content, analyze performance, identify important comments, draft replies in their own voice, and receive personalized recommendations.

Cate Blanchett launches ‘Human Consent Registry’ to help protect people from AI likeness misuse

Academy Award-winning actress Cate Blanchett has unveiled the Human Consent Registry, a free platform that lets people specify whether AI companies can use their name, image, voice, and other personal characteristics. Users can choose different consent levels, receive a unique Human Consent ID, and create a machine-readable record of their preferences.

Europe pushes back against proposed U.S. chip export restrictions

The Dutch government is urging U.S. lawmakers to reconsider the proposed MATCH Act, arguing that the legislation would significantly impact ASML, the Netherlands-based semiconductor equipment maker that supplies the world's most advanced chipmaking lithography machines.

Samsung reportedly plans $648 billion investment to expand South Korea’s AI infrastructure

Samsung Group is reportedly preparing to invest 1,000 trillion won ($648 billion) in South Korea over the next decade as part of a national initiative to strengthen the country’s AI and semiconductor leadership. The proposed investment would support AI data centers, semiconductor manufacturing, batteries, and display technologies.

OpenAI mulls delaying IPO until 2027

OpenAI is reportedly considering postponing its initial public offering until 2027 to pursue a $1 trillion valuation, rather than going public sooner at a lower valuation.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

In partnership with

Welcome to another edition of Horizon AI,

Google recently began rolling out a new privacy setting that many users have overlooked. In today's issue, we'll show you what it does and how to turn it off if you'd rather keep your data from being saved and used to improve Google's services and AI models.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• OpenAI Introduces Its First Custom Chip, Developed with Broadcom

• Report says Grok's Traffic Is Mostly Driven By Adult Content

• AI Tutorial: A New Google Privacy Setting Is Rolling Out — Here's How to Turn It Off

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Introduces Its First Custom Chip, Developed with Broadcom

OpenAI has introduced Jalapeño, its first custom-built AI inference chip, designed to power its inference workloads more efficiently and reduce the company's reliance on Nvidia GPUs.

Details:

• Jalapeño is purpose-built for inference, the process of running trained AI models to generate responses to user requests.

• The “Intelligence Processor” was designed and manufactured in collaboration with Broadcom, with OpenAI stating that its own AI models assisted in the chip's development process.

• While the chip remains in testing, OpenAI says early benchmarks show significantly better performance per watt than current state-of-the-art alternative.

• OpenAI emphasized the chip's low operating cost when running real-time coding models, but more performance-intensive tasks, like pre-training, are still expected to rely largely on Nvidia hardware.

OpenAI described Jalapeño as the "first AI accelerator in a multi-generation compute platform the companies are building together to make advanced AI faster, more reliable, and more accessible to more people." The companies are aiming for initial deployment by the end of 2026, "expanding in the years ahead."

Source

TOGETHER WITH CUEY

ChatGPT and Claude often disagree. Cuey shows you when.

You ask ChatGPT something important. It answers with total confidence. Ask Claude or Gemini the same thing, and the answer changes. One of them is wrong, and you are about to act on it.

Cuey runs your prompt past ChatGPT, Claude, and Gemini in a single tab, so you spot any disagreements before they cost you. Your memory and context follow you across all of them, no starting over. It works beside the tools you already use, with Grok, Mistral, and more than 30 others.

Free on Chrome, nothing to set up.

XAI

Report says Grok's Traffic Is Mostly Driven By Adult Content

A new report claims that Elon Musk's xAI is leaning into explicit content generation as a core driver of Grok's traffic, with NSFW content now accounting for the majority of the platform's activity.

Details:

• According to The Information, two former xAI employees estimate that well over half of all Grok traffic goes to NSFW images, videos, roleplay chats, or other adult content. Even Grok's coding model sees frequent requests of this nature.

• Vital Knowledge analyst Adam Crisafulli described the strategy as "a desperate attempt for relevancy," arguing that xAI has fallen further behind leading AI developers.

• One notable controversy earlier this year involved Grok's image tools being used to "undress" photos of real people without their consent, forcing the company to introduce measures to prevent further misuse.

Grok was launched by Elon Musk as a direct competitor to ChatGPT, Claude, and Google Gemini, with a focus on being a less restrictive AI assistant. However, the platform has struggled to keep pace with leading frontier models, with its web traffic declining this year while competitors like Claude (+369%) and Gemini (+40%) have seen significant growth over the same period, according to Similarweb.

Source

AI Tutorial

A New Google Privacy Setting Is Rolling Out — Here's How to Turn It Off

Did you get this email? Google started sending everyone this email last week about a new setting.

The new setting controls whether media used with supported Google search services (such as images submitted through Google Lens) will be used to improve the company's services and AI models. Google says the data eventually gets stripped of your identity once it's used for training, but they don't say when, and the collection is already happening.

How to turn it off:

1. On your phone or computer, visit: http://myactivity.google.com/activitycontrols

2. If you see a section called Search services history, tap into it and turn off “Save media.”

3. If that section isn't there yet (which is the case for most people right now, as it's rolling out gradually), go to Web and app activity and uncheck Voice, audio & visual search history as a stopgap.

You can then set a reminder on your phone to go back in three or four weeks and confirm the Search services history toggle is off once it appears on your account, as Google says it is "rolling out gradually over the next few months."

AI Tools to check out

🔥 Plotly Cloud: Transform any dataset into a data app in minutes.

✅ Exa Connect: Connect your agents to the world's public and private data.

📞 AgenticCalling: Give your AI the power to make phone calls.

👉 Propane: Automatic customer context for product teams and agents.

🌐 Aside: The first AI browser built to do real work for you. It works across your logged-in websites and handles complex work other agents can't.

AI Findings/Resources

💂 Why U.S. AI giants like Anthropic, OpenAI are launching major expansions in London

🗣️ OpenAI's upcoming Bidi 1 voice model will be able to translate in real time

🛑 Companies are scrambling to stop employees from maxing out AI budgets with small tasks

📌 Anthropic engineer shares Claude Tag best practices to help you get the most out of Claude's new Slack integration

🤯 How GLM-5.2 beat Fable 5 at website design

The latest in AI and Tech

OpenAI rolls out updated version of GPT-5.5 Instant

The company says the model is now better at understanding the intent behind a question, handles complex constraints more reliably, and delivers more useful shopping and local recommendations.

More Google AI researchers leave the company

Top AI researchers Jonas Adler and Alexander Pritzel are leaving Google for Anthropic, according to Bloomberg.

Senior research engineer Arthur Conmy also announced his departure from Google to join Anthropic, "to work on aligning upcoming models as they're trained."

Figma adds code layers, animations, and more AI features

The design platform has unveiled a major update introducing code layers, built-in support for animations, and the ability to create custom plug-ins for various tasks using AI.

Google introduces computer use in Gemini 3.5 Flash

Google has integrated computer use directly into Gemini 3.5 Flash, allowing developers to build AI agents that can interact with websites, desktop apps, and mobile interfaces.

OpenAI and Microsoft face lawsuit from local newspapers

A coalition of nearly 400 local newspapers has filed a lawsuit against OpenAI and Microsoft, accusing the companies of “scraping, copying, and ingesting” their journalism without permission or compensation to train AI models.

The case adds to a growing wave of legal challenges against OpenAI, which already faces lawsuits from major outlets including The New York Times, Ziff Davis, Merriam-Webster, and Encyclopedia Britannica.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

Need a realistic voiceover without recording your own audio? In today's issue, we'll show you how to use AI to turn any text into high-quality speech, making it easy to create narration for videos, presentations, podcasts, and more.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Nvidia Says Next-Generation AI Data Centers Could Nearly Eliminate Water Use

• Meta Debuts AI-Powered Meta Glasses, Starting at $299

• AI Tutorial: Turn Text into Natural-Sounding Speech with AI

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

NVIDIA

Nvidia Says Next-Generation AI Data Centers Could Nearly Eliminate Water Use

Nvidia says its upcoming Rubin AI data center reference design has "eliminated massive amounts of power usage and pretty much all water usage" by replacing traditional cooling systems with a fully liquid-cooled architecture that operates at higher temperatures.

Details:

• Nvidia claims switching to 100% liquid cooling will eliminate the need for conventional air-cooling systems that consume large amounts of water.

• According to the company, the new design can reduce water usage “from roughly 2.6 million gallons per megawatt per year for conventional cooling-tower-based systems to near zero — up to a 100 percent reduction.”

• The efficiency gains come from allowing AI servers to operate at significantly higher temperatures (up to 113°F / 45°C).

• With the new system, "heat is captured directly at the chip and transported through liquid loops operating at much higher temperatures, allowing outdoor dry coolers to reject heat efficiently for much of the year."

Nvidia did not disclose the additional construction costs compared with conventional air-cooled facilities.

Source

TOGETHER WITH CASH APP

Set it, forget it, and save for your goals

Make saving automatic with Cash App. Round up your spare change from every purchase, earn up to 3.25% interest, and transfer money between your balances whenever you want—all with no hidden fees or minimum balance requirements.

Make unlimited transfers between your Cash and Savings balances
Keep your money safe with 24/7 fraud monitoring and built-in security features

Saving is easier with Cash App

Start saving with Cash App

Cash App is a financial services platform, not a bank. Banking services provided by Cash App’s bank partner(s). Prepaid debit cards issued by Sutton Bank, Member FDIC. See Terms & Conditions. Eligibility restrictions apply to some benefits. See Terms and Conditions for more information.

META

Meta Debuts AI-Powered Meta Glasses, Starting at $299

Meta has unveiled a new line of Meta Glasses, introducing a more affordable range of AI-powered smart glasses starting at $299.

Details:

• While the devices continue to be manufactured in partnership with EssilorLuxottica, they are the first to carry Meta's own branding instead of the Ray-Ban or Oakley names.

• The glasses do not include a display. Instead, they feature a built-in camera for capturing photos and videos, open-ear speakers for audio playback and calls, and a dedicated hardware button that launches the Meta AI assistant by default, with the option to customize it for other functions.

• Powered by Meta AI, the glasses can answer questions, provide sports scores, recommend nearby restaurants, interpret what the wearer is looking at through the camera, and assist with everyday tasks using voice commands.

The launch comes amid intensifying competition in the smart glasses market, though according to Counterpoint Research, Meta and EssilorLuxottica still control more than 80% of the global smart glasses market.

Source

AI Tutorial

Turn Text into Natural-Sounding Speech with AI

You can transform text into natural-sounding speech in seconds with the Kokoro Text-to-Speech model:

1. Go to Kokoro-TTS page on Hugging Face.

2. Choose your desired voice and enter your text. Be mindful of punctuation to get better results.

3. Click on "Generate" and wait a few seconds. You can experiment with changing the speed and voices until you get your desired result.

4. Use the Generate tab for up to ~500 characters, or switch to Stream for up to 5,000 characters.

5. Share or download your generated audio.

AI Tools to check out

📞 Caddie: An AI copilot that joins your sales calls. You talk — it researches, recalls, and answers, live, while you're still on the call.

✅ Clarafy: Translate chaotic brain-dumps into perfectly formatted text in a single click.

🔊 Cleanaudio: Remove background noise from audio and video with AI.

🦾 Vokal: A collaboration space where teammates and AI agents work from shared channels, tasks, docs, tools, memory, and Knowledge Base.

👉 Franz: All-in-one messaging app for Mac, Windows & Linux.

AI Findings/Resources

🔮 Cloudflare CEO Matthew Prince predicts: "I think AI is going to destroy small businesses."

📱 Beyond Siri: Here are the practical AI features coming to your iPhone in iOS 27

🧠 How to build an AI second brain with Claude and Obsidian that gets smarter every day

🤔 How to survive AI mass replacement (& escape wage slavery)

🚀 “AI for impact” is thriving in Asia – so how can public systems speed up adoption?

The latest in AI and Tech

Krea 2 Raw and Krea 2 Turbo are now available as open-weight models

Turbo is designed for fast, polished text-to-image generation, while RAW serves as a malleable base checkpoint ideal for LoRA training in ComfyUI and beyond.

OpenAI launches "Patch the Planet"

The new initiative aims to help open source maintainers secure their projects. The program combines human security experts with OpenAI’s AI tools to reduce the burden on maintainers and strengthen the security of widely used open source software before bugs can be exploited.

US presses Meta to submit AI models for review

The Trump administration has reportedly asked Meta to voluntarily submit its AI models for government review, allowing officials to evaluate their capabilities and potential security risks.

Five Eyes alliance warns frontier AI cyber threats are ‘months’ away

The Five Eyes intelligence alliance has warned that frontier AI models could dramatically reshape offensive and defensive cyber operations within months, not years. In a joint statement, the signals intelligence agencies of Australia, the US, the UK, New Zealand, and Canada are urging leaders in business and politics to "act now."

Samsung deploys ChatGPT Enterprise and Codex to employees

The company is rolling out ChatGPT Enterprise and Codex to all employees in South Korea and its Device eXperience (DX) division worldwide in one of OpenAI’s largest enterprise partnerships to date.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

AI features like writing assistants, photo editing, and voice commands have become a staple on modern smartphones, but they can quietly drain your battery in the background. In today's issue, we'll show you how to find out if AI is affecting your battery life and what you can do about it, on both iPhone and Android.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• ByteDance Announces Seedance 2.5 Along With an Updated Seedance 2.0

• SpaceX Signs $6.3 Billion Data Center Deal with AI Startup Reflection AI

• AI Tutorial: How to Check If AI Is Draining Your iPhone or Android Battery

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

BYTEDANCE

ByteDance Announces Seedance 2.5 Along With an Updated Seedance 2.0

ByteDance has introduced Seedance 2.5, its latest AI video generation model, featuring significantly longer video generation, expanded creative controls, and new editing capabilities.

Details:

• The model was announced at Volcano Engine's FORCE conference in Beijing and is scheduled to launch in China in early July. An international release timeline has not yet been announced.

• Seedance 2.5 can generate continuous video clips up to 30 seconds long, without requiring multiple clips to be stitched together. The generated videos can include scene transitions, pacing changes, and more complex narratives within a single output.

• The model expands user control by allowing prompts to include up to 50 reference assets (including images, videos, and audio files) compared to 12 reference inputs supported by Seedance 2.0. More reference materials give users more precise control over the videos they want to create.

• Seedance 2.5 also introduces post-generation editing, allowing users to modify AI-generated videos while keeping the visual style and look intact.

Alongside Seedance 2.5, the company announced Doubao 2.1 Pro, a large language model that ByteDance claims delivers high-end performance at roughly 80% lower cost than Anthropic's Claude Opus 4; Seedream 5.0 Pro, an upgraded image generation model; Seed-Audio 1.0, a new AI audio generation model; and an update to Seedance 2.0, which now supports native 4K video generation with 10-bit color depth.

Source

TOGETHER WITH LITTLEBIRD

You don't have to carry the whole business in your head.

Littlebird remembers every doc, call, and decision so you stop re-explaining context to ChatGPT. Ask a question, get the answer back with the receipts. No long prompts. Encrypted end to end, your data stays yours. Get $20 of Littlebird credit on signup.

Signup here

SPACEX

SpaceX Signs $6.3 Billion Data Center Deal with AI Startup Reflection AI

SpaceX has signed a major long-term infrastructure agreement with Reflection AI, a startup developing open-source AI models.

Details:

• Under the agreement, Reflection AI will lease data center capacity from SpaceX to power the training and deployment of its AI models. The contract is expected to generate $150 million in monthly revenue for SpaceX beginning July 1, totaling roughly $6.3 billion if it runs through its full term.

• The contract extends through the end of 2029, although either company may terminate the agreement with 90 days' notice after the initial three-month period.

• Backed by Nvidia, Reflection AI is reportedly valued at $25 billion and aims to compete with the growing number of advanced open-source AI models emerging from China.

• A Reflection AI spokesperson said the additional computing capacity will provide the infrastructure needed to continue scaling its models, stating that "more compute means more runway to build the world's best open models at scale."

The deal is among the first for SpaceX since its public market debut earlier this month, which valued it at $1.77 trillion. Reflection has previously signed similar agreements with Anthropic and Google.

Source

AI Tutorial

How to Check If AI Is Draining Your iPhone or Android Battery

First, Check If AI Is Actually the Problem

Look at your phone's battery usage report to see what's consuming the most power.

On iPhone:

• Open Settings

• Tap Battery

On Android:

• Open Settings

• Tap Battery

• Select Battery Usage

If you notice AI assistants, photo apps, productivity tools, or other AI-powered services near the top of the list, they're likely contributing to your battery drain. Often, it's not one feature but several AI services running together in the background.

iPhone Settings Worth Reviewing

If you're using an iPhone, these are the biggest AI-related battery drains to check:

• Apple Intelligence: If you rarely use AI writing tools, summaries, or image generation, consider disabling the features you don't need instead of leaving everything enabled.

• Siri Suggestions: Go to Settings > Siri and review the Suggestions options. Turning off suggestions you never use can reduce background activity.

• Background App Refresh: Open Settings > General > Background App Refresh and disable it for AI-powered apps you don't need updating constantly.

Android Settings Worth Reviewing

Android phones also include several AI features that may be running behind the scenes.

• Voice Activation: If you don't regularly use Gemini or "Hey Google," review these settings by going to Settings > Google > Search, Assistant & Voice and disable voice activation if you don't need it.

• AI Photo Processing: Many Android phones use AI to scan and sort your photos automatically. Check your photo app settings and decide whether automatic photo analysis, categorization, cloud syncing, or advanced image processing are features you actually use.

• Battery Optimization: Open your battery settings and review Battery Optimization, App Battery Usage, or Background Restrictions. These tools can help identify power-hungry apps that continue running in the background.

AI Tools to check out

🔍 Latitude: Open-source AI agent monitoring platform. Capture agent trajectories, discover behavior patterns, and catch issues before your users do

👉 Folk: A personal AI that lives in your iMessage, Telegram, and WhatsApp. It tracks flights, books dinners, and keeps you connected with your friends.

👨‍💻 Vibecode: Turn your app idea into reality in minutes.

📓 NexoMind: A private AI journaling platform built to help people stop overthinking, process emotions, and gain mental clarity through structured AI reflection.

✨ Resumma: Generate AI summaries, translate text and images instantly.

AI Findings/Resources

📚 Notion Workers – Full tutorial

🗣 A demo of the new ChatGPT voice mode shows it can listen and speak at the same time, allowing for a more natural conversation

📌 6 tips for prompting Lyria 3 (AI music generator) in the Gemini app

👀 A Reddit user claims ChatGPT fixed their 9-year chronic pain

🤨 Google is reportedly testing a new reCAPTCHA system that replaces image puzzles with simple hand gestures, as bots and AI tools have gotten better at solving them

The latest in AI and Tech

Anthropic launches Claude Tag for Slack

The new Slack integration lets users assign tasks to Claude by mentioning @Claude in a Slack channel. The AI agent can write and merge pull requests, find sales data, analyze information, take on delegated tasks, and more.

Google Sheets gets AI formula fixes with Gemini

Google is rolling out a new Gemini-powered "Fix" button in Google Sheets that can automatically diagnose and repair broken formulas. When you click a cell with a formula error, Gemini suggests a fix in the sidebar, explaining the error and providing the right formula.

Meta pauses employee AI tracking tool

Meta has paused an internal AI training program after an internal leak revealed that employees across the company could potentially access sensitive data, including private conversations, performance information, and transcriptions. While Meta says it has no evidence of improper access, it has suspended the program as it investigates the issue.

Oracle cuts 21,000 jobs as it sharpens focus on AI

The company had a total workforce of 141,000 as of May 31, 2026, compared with about 162,000 as of the same period last year, according to its ​annual report released on Monday.

"The deployment of AI technologies across our operations has resulted, and may continue to result, in reductions to our workforce," the report says.

ElevenLabs launches AI tool to auto-translate Ad campaigns

ElevenLabs has introduced Ads Engine, a new AI-powered tool that automatically localizes ad campaigns across 50+ languages. It can translate text, dub videos, adapt images, and publish localized creatives to platforms like Google Ads and Meta Ads.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we'll show you how to use a single prompt to generate polished, encyclopedia-style infographics that organize complex information into clear, visually engaging reference cards in just a few minutes, perfect for studying, sharing, or creating your own collection of knowledge cards.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Japanese AI Firm Sakana AI Releases 'Fugu'

• Google Invests $75 Million in A24 to Develop AI Tools for Filmmakers

• AI Tutorial: Create Encyclopedia-Style Infographics with AI

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

SAKANA AI

Japanese AI Firm Sakana AI Releases 'Fugu'

Japanese AI startup Sakana AI has unveiled Fugu, a new AI system that dynamically orchestrates multiple language models to deliver frontier-level performance while reducing reliance on any one AI provider.

Details:

• Fugu is a language model trained to select, coordinate, and combine multiple LLMs from a swappable agent pool.

• It behaves like a single model: users send a request to one endpoint, and Fugu decides how to handle it, solving it directly when that is enough or assembling a team of expert models when the task calls for more.

• Sakana AI is releasing two versions: Fugu for low-latency everyday tasks (coding, code review, conversational AI) and Fugu Ultra for complex multi-step reasoning (AI research, scientific paper reproduction, cybersecurity analysis, patent searches, and technical literature review).

• Both models can be accessed via a single OpenAI-compatible API.

• The company claims Fugu Ultra performs on par with Anthropic's Fable 5 and Mythos Preview across coding, reasoning, and science benchmarks, and outperforms models like Gemini 3.1 Pro, Opus 4.8, and GPT-5.5 on tasks like AutoResearch, mechanical design, and financial forecasting.

The company argues its architecture also reduces vendor lock-in, letting organizations swap models if a provider becomes unavailable due to policy changes, export controls, or service disruptions. It cites recent U.S. export restrictions as a reason to avoid single-provider dependency for mission-critical applications.

Source

TOGETHER WITH GROUNDFLOOR

Over the Market Volatility? Smart Investors Are Moving Into Fixed Income That Earns 4x More Than CDs.

While stocks swing and bond yields compress, Groundfloor Notes quietly pay up to 8.25% fixed — no market correlation, no daily price swings, no surprises. Just a locked rate backed by diversified, first-lien real estate loans professionally underwritten and managed by Groundfloor. Since 2013, over 300,000 investors have trusted Groundfloor for SEC-qualified investment offerings.
Choose a 30-day, 90-day, or 12-month term and know exactly what you'll earn before you invest. That's a yield most fixed-income investors haven't seen in years, with the flexibility to pick a timeline that fits your strategy. Since launching Notes in 2018, Groundfloor has made every scheduled principal and interest payment at maturity. Every investor has seen positive returns — through rate hikes, inflation, and market corrections.
For a limited time, get a $100 bonus added to your account when you invest $1,000+. Today's 8.25% rate won't last in this environment — lock it in and claim your bonus now.

Invest With Groundfloor Notes Now

GOOGLE

Google Invests $75 Million in A24 to Develop AI Tools for Filmmakers

Google has invested approximately $75 million in independent film studio A24 as part of a strategic partnership with Google DeepMind to develop AI-powered filmmaking tools.

Details:

• The partnership gives A24 access to Google DeepMind's AI research, infrastructure, and engineering expertise, with researchers from both organizations collaborating to build new tools tailored for film production.

• The agreement does not grant Google access to A24's film library, intellectual property, or production data.

• A24 says the collaboration will focus on AI tools that support filmmakers throughout the creative process while preserving artistic control, rather than emphasizing cost reduction or fully AI-generated content.

• One of the first areas of development is AI-assisted storyboarding, enabling filmmakers to rapidly visualize scenes and production concepts before filming. A24 executives describe the initiative as a way to streamline pre-production without replacing creative decision-making.

The agreement follows a broader trend of collaboration between Hollywood studios and AI companies. Recent examples include Lionsgate's expanded partnership with Runway AI, Netflix's acquisition of filmmaker-focused AI startup InterPositive, and ongoing experimentation across the entertainment industry with AI-assisted production tools.

Source

AI Tutorial

Create Encyclopedia-Style Infographics with AI

Source: @oggii_0 on X

Go to ChatGPT, select 'Create Image,' and use this prompt:

Based on { TOPIC }, create a high-quality vertical “encyclopedia-style educational infographic image.”

This image should NOT look like a regular poster or a simple illustration. Instead, it should feel like a structured knowledge guide that combines:

the feeling of a collectible reference handbook

a modern encyclopedia page

a lifestyle knowledge card

and a highly shareable social-media infographic

The overall style should resemble a premium natural-history guidebook mixed with modern editorial infographic design.

The image should include:

One beautiful and highly detailed main subject image

Several zoomed-in detail sections highlighting important features

Multiple modular information panels with rounded corners

Clear title hierarchy and highlighted key labels

Concise yet information-rich encyclopedia content

Visual scoring systems, quick summaries, or “Top 5” modules

The information sections should automatically adapt to the topic. Select and combine relevant categories such as:

Basic profile

Classification / taxonomy

Physical characteristics

Behavior / ecology / habits

Structure or formation mechanisms

Growth conditions or usage methods

Care, maintenance, or optimization tips

Risks, warnings, and important notes

Suitable users or application scenarios

Pros and cons comparison

Quick rating tags or summary cards

Visual requirements:

Clean light-colored background

Soft and elegant color palette

Gentle shadows

Small refined icons

Rounded information boxes

Organized editorial layout

High information density without feeling crowded

Comfortable reading experience

The final result should feel like a real publishable encyclopedia knowledge card designed for reading, collecting, and creating as part of a consistent series — NOT like a commercial advertisement poster.

The image must strongly emphasize:

“knowledge integration + modular information design + handbook/reference-style presentation.”

AI Tools to check out

🌎 LanguageTool: An AI-based grammar checker. Paste your text or start typing below to check grammatical errors and spelling mistakes across languages.

🔍 SciSpace: The AI research assistant for academics. Run systematic literature reviews on 280M+ papers, and write papers with cited sources.

✅ Cekura: Test, monitor and self improve voice & chat AI agents.

👨‍💻 Cubic: AI code reviews for complex codebases

📹 Vmake: An all-in-one UGC video generator that helps small businesses and creators generate, enhance, and optimize high-performing videos

AI Findings/Resources

✨ This Korean guy used AI video to make ads for products that should exist

👓 I tested AI glasses in Paris. Here’s what they got wrong

⚔️ Sakana Fugu Ultra vs. Claude Opus 4.8, same prompt

👀 Loops explained: Claude, GPT, Mira and what actually works

🛡️ Google DeepMind releases its AI Control Roadmap, 'a defense-in-depth system to securely manage advanced, potentially misaligned AI agents.'

The latest in AI and Tech

Google's Nobel Prize-winning AI researcher joins Anthropic

John Jumper, who shared the 2024 Nobel Prize in Chemistry for co-developing AlphaFold, is leaving Google DeepMind after nearly nine years to join Anthropic. His departure comes shortly after Gemini co-lead Noam Shazeer's move to OpenAI, continuing a string of high-profile exits from Google.

Amazon drops Sam Altman movie after announcing OpenAI partnership

Amazon MGM Studios has dropped Artificial, a film about OpenAI CEO Sam Altman's brief ouster in 2023, despite the project being nearly complete. The decision comes months after Amazon signed a major partnership with OpenAI. The movie is now being shopped to other studios.

AI helps identify rare disease diagnoses for 18 children at Boston Children's Hospital

Researchers at Boston Children’s Hospital used OpenAI’s o3 model to identify 18 new diagnoses among 376 patients with rare diseases who had previously gone undiagnosed despite extensive genetic analysis. Researchers emphasized that human experts still reviewed all findings and that AI serves as a diagnostic aid rather than a replacement for medical professionals.

Norway restricts AI use in schools

Norway will largely ban generative AI tools in elementary schools and limit their use in secondary schools starting this academic year, citing concerns that excessive AI use undermines core skills such as reading, writing, and math. Under the new policy, students in grades 1–7 will generally not be allowed to use AI, while older students may use it under teacher supervision.

Getty Images signs multi-year licensing deal with OpenAI

Getty Images has signed a multi-year licensing agreement with OpenAI that will bring licensed photos from Getty's catalog to ChatGPT's search and discovery features. The companies did not disclose financial terms or whether the licensed content will be used to train future AI models.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we take a look at a conversation with Wharton professor and bestselling author Ethan Mollick, exploring the practical reality of living and working with AI. Mollick shares grounded, actionable strategies for navigating a rapidly shifting workforce, why human judgment and experience matter more than ever, and how we can use these tools to enhance our thinking rather than replace it.

Let’s jump right in!

Read Time: 4.5’ min

Here's what's new today in the Horizon AI

• Chart of the week: Youngest Billionaires

• Americans Embrace AI, But with a Skeptical Eye

• AI Findings/Resources

• AI tools to check out

• Video of the week

TOGETHER WITH STORECLAW

Stop Paying for 6 Tools. One AI Does It All

Most e-commerce sellers are running their store across 6 to 8 separate tools — and paying hundreds of dollars a month for the privilege. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.

It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.

Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.

Free to start. No credit card required.

Try it for free today

Chart of the week

Youngest Billionaires

• According to Forbes' 2026 Billionaires List, 35 people under the age of 30 make the cut, representing about 1% of the world's billionaires.

• Of the 35, most inherited their fortunes, but 12 are self-made, having built their wealth primarily in AI, tech, and prediction markets.

• The youngest among them are Surya Midha, Brendan Foody, and Adarsh Hiremath, the trio behind AI recruiting startup Mercor.

• Other notable names include Sweden's Fabian Hedin, 26, who co-founded AI coding startup Lovable, and Cursor founders Michael Truell (25), Aman Sanger (25), and Arvid Lunnemark.

Source

AI News

AI RESEARCH

Americans Embrace AI, But with a Skeptical Eye

A new Pew Research Center survey shows that while AI chatbot usage continues to grow rapidly across the United States, public skepticism remains high, with nearly two-thirds of Americans believing the technology is advancing too quickly.

Details:

• 49% of Americans now report using AI chatbots at least occasionally, up significantly from 33% in 2024.

• Despite increased adoption, 63% of Americans believe AI technology is advancing too quickly.

• Although younger respondents were most likely to have tried AI, it is actually adults between the ages of 30 and 49 who use it the most, with 34% saying they use AI chatbots at least once per day. The survey suggests this may be driven in part by workplace adoption.

• Approximately 40% of Americans report using AI for work-related tasks. Among respondents, 30% say AI improves their productivity, while 28% believe it helps them stay better informed.

Public sentiment remains largely cautious, with only 16% of respondents believing AI will have a positive impact on society. Younger Americans are the most skeptical, with 48% of adults aged 18–29 believing AI will have a negative impact on society, compared to just 14% who expect a positive one. In contrast, older age groups tend to use AI less, but hold a more favorable or neutral view of it.

Source

AI Findings/Resources

😐 Amazon investigating engineers who criticized AI data center expansion

👀 Who are SpaceX’s billionaire shareholders?

🔮 Elon Musk says smartphones won't exist in 5 years

🧹 Why an AI company cleaned my New York City apartment for free

💼 Jeff Bezos says AI will create more jobs for humans, not replace them

AI Tools to check out

👉 Dola AI: AI chat assistant for intelligent conversations, writing, translation and programming.

📢 Okara Influencer: The world's first influencer Agent. Tell it what you want to promote and it finds creators, reaches out, manages campaigns, handles payments, and gets content live.

🌐 Framer: AI tools to design, build, and improve websites.

🦾 IvaBot: AI SEO tool that finds and fixes SEO issues, discovers niche keywords, and writes content that people read, Google ranks, and AI tools cite.

✅ Tictable: A minimalist data studio for the AI age.

Video of the week

The AI Skills Nobody is Teaching (And Everyone Needs)

In this conversation, Mollick shares grounded, actionable strategies for navigating our rapidly shifting workforce. They discuss how AI is dismantling traditional career ladders, why human "taste" and experience are becoming the ultimate competitive advantages, and how we can intentionally use these tools for cognitive augmentation rather than outright automation.

The Practical Middle Ground

Mollick consciously rejects both the "doomer" and "zealot" camps of the AI debate. Instead, he frames AI as a general-purpose technology (akin to the internet) that is deeply woven into human systems. He emphasizes that rather than being passive bystanders, we maintain an immense amount of agency over how these tools are personally and professionally integrated.

The Breaking of the Talent Pipeline

For millennia, specialized skills were passed down through apprenticeship and "grunt work" evaluated by middle management. Mollick warns that this foundational structure has broken. Because current AI models outperform entry-level workers at basic tasks, managers are tempted to delegate solely to machines, threatening the survival of traditional corporate talent pipelines.

Commoditization and the Premium on Taste

As AI models like Gemini, ChatGPT, and Claude commoditize high-quality work, standing out in the job market will no longer depend on basic technical proficiency. When anyone can generate flawless code or clean copy, a human's unique competitive edge shifts entirely to their personal taste, vision, and distinct point of view.

Practical Strategies for Cognitive Augmentation

Mollick shares explicit, high-value tactics to instantly level up your interactions with modern thinking models:

• The Power of Persona Demands: Don't just ask the AI for a generic critique. Instruct it to review your ideas from highly specific, varied perspectives, such as a harsh social media critic looking for flaws, an industry novice, or a specific professional archetype.

• Forcing the Critic: Naturally polite, AI defaults to agreeing with your inputs. To push your thinking further, explicitly command the model to act as a harsh critic and challenge your underlying assumptions.

• The Style Guide Trick: To bypass generic "AI voice," feed the tool a large sample of your authentic writing. Ask it to analyze and write a two-page summary detailing your tone, vocabulary, and structural habits, then paste those rules into your custom instructions.

A Shift in Job Burdens

Jobs are composed of many interconnected tasks. Mollick explains that AI rarely automates a whole job out of existence; instead, it eliminates specific operational bottlenecks [31:10]. In writing and coding, the heavy lifting is shifting away from raw production toward high-level architecture, rigorous editing, and rapid evaluation.

Personalized Education: The Holy Grail

While structural shifts create widespread organizational chaos, Mollick highlights a profound silver lining in education: personalized AI tutoring [48:34]. Armed with an effective "theory of mind," these systems can dynamically adapt to an individual student's learning speed, meet them at their current cognitive level, and generate hyper-tailored analogies based on their specific personal interests.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we're taking a look at OpenAI's "Scheduled Tasks," a feature that just got a major update with a dedicated Scheduled page, improved task creation and editing flows, and more flexible scheduling options.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Adobe Rolls Out AI Assistants Across Photoshop, Premiere, Illustrator, and More

• Perplexity Launches 'Brain,' a Self-Improving Memory System for Its AI Agent

• AI Tutorial: How to use ChatGPT's upgraded Scheduled Tasks

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

ADOBE

Adobe Rolls Out AI Assistants Across Photoshop, Premiere, Illustrator, and More

Adobe is updating its AI assistant and adding it to Photoshop, Premiere, Illustrator, InDesign, and Frame. io.

Details:

• Each application now includes a dedicated AI assistant powered by Adobe's Conversational Creative Agent, with every assistant specialized for the workflows and tools of its respective app rather than functioning as a single universal chatbot.

• Users can interact with the assistants through natural language prompts to automate repetitive tasks, organize projects, and perform multi-step editing operations without manually navigating complex menus or workflows.

• In Photoshop for example, you can organize layers, swap backgrounds, resize assets for different platforms, and more, simply by "describing the desired outcome."

According to Adobe, the goal is to provide creators with an AI assistant in every major Creative Cloud app that can understand the context of their projects and help execute complex workflows while leaving creative direction and final decisions in the hands of users.

Source

TOGETHER WITH METRO BY T-MOBILE

Free Samsung Galaxy A17 5G

You can currently get the new Samsung Galaxy A17 5G for FREE at Metro by T-Mobile when you bring your number and sign up for a qualifying unlimited plan. This deal includes no activation fees, AND a 5-year price guarantee on your service. Taxes and Fees are included.
The Galaxy A17 5G is built for longevity and smart performance. It is the first in its class to offer six years of guaranteed OS and security updates. Helpful AI tools like Circle to Search and Gemini, bring flagship level-convenience to an accessible price point.
Switching to Metro has never been easier, just bring your number and get the Latest Samsung Galaxy A17 5G for FREE.

Choose Your Metro by T-Mobile Plan and Get Your Free Samsung Galaxy A17 Phone

Terms & Conditions apply. Not available if with Metro in the last 180 days. If using >35GB/mo. May notice reduced speeds.

PERPLEXITY

Perplexity Launches 'Brain,' a Self-Improving Memory System for Its AI Agent

Perplexity has introduced Brain, a new memory layer for its Computer AI agent that learns from previous tasks allowing the agent to become more efficient over time.

Details:

• Unlike traditional AI memory that primarily stores user preferences, Brain focuses on the agent's work history. It records completed tasks, successful workflows, failed attempts, user corrections, and the sources or connectors used to accomplish each job.

• Every completed task is added to a context graph, which links memories back to the original sessions, files, or sources. At scheduled intervals (overnight by default) Brain summarizes this information into a personal knowledge base that is loaded before future tasks, enabling the agent to begin with accumulated context instead of starting from scratch.

• Perplexity says this persistent memory allows Computer to reuse effective workflows, avoid previously unsuccessful approaches, and provide full transparency by letting users trace every stored memory back to its origin.

• According to Perplexity's internal testing, Brain improves answer accuracy by 25% on recurring tasks, increases recall by 16%, and reduces the computational cost of context-heavy workloads by 13%.

Brain is available in Research Preview for Perplexity Max and Enterprise Max subscribers. Perplexity says additional capabilities are planned, though it has not announced a timeline for future releases.

Source

AI Tutorial

How to use ChatGPT's upgraded Scheduled Tasks

1. Go to ChatGPT and sign in.

2. Look for the Scheduled page in the ChatGPT sidebar on web or mobile.

3. Create a new task and choose when it should run.

4. You can also create a new task by simply asking ChatGPT in a new chat. For example: "The day before every Portugal match, send me a scouting report on the opponent including key players and tactics."

You can schedule one-off and recurring tasks, and also ask ChatGPT to check for changes and notify you when there's a meaningful update.

5. Go to the Scheduled page to manage your tasks.

Note: The new Scheduled Tasks experience is available to Go, Plus, Pro, Business, and Enterprise users on web and mobile.

AI Tools to check out

🔥 Fastlane: Just enter your website and it creates thousands of viral videos promoting your product in one click.

📚 StudyPal: The #1 tool to learn anything 10x faster.

✅ Bond: It connects to your tools, learns how your company works, and turns scattered tasks into a self-managing to-do list that always knows what you need to do next.

🌎 HappyOyster 1.0: An open-ended world model product for real-time world creation and interaction

📽 Ponder: A powerful agentic video editor where creative AI and humans collaborate to tell world-class stories.

TOGETHER WITH STORECLAW

Stop Paying for 6 Tools. One AI Does It All.

Most e-commerce sellers juggle 6–8 tools and pay hundreds monthly to keep operations running. StoreClaw replaces the stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks profit 24/7. Connect your store and let AI handle the work — no prompts, no complex setup, no credit card required.

Try if for free today

AI Findings/Resources

📚 Full course: Build your own OpenClaw-like AI agent from scratch using Vercel, Composio, and Supermemory

👀 You can now teach Codex by demonstration

👉 AI as Normal Technology: An alternative to the vision of AI as a potential superintelligence

⚽ FIFA uses robots to patrol during World Cup (not for face scanning)

🤯 Z .ai founder is confident that they can make a fable-class GLM model before the end of the year

The latest in AI and Tech

OpenAI introduces LifeSciBench

The new benchmark, developed with input from 173 scientists across the biotechnology and pharmaceutical industries, is designed to measure how well AI models support real-world life science research.

The company also announced that it has upgraded ChatGPT's healthcare capabilities with the GPT-5.5 Instant model, which now matches the performance of more expensive reasoning models on health-focused benchmarks such as HealthBench and HealthBench Professional.

Roelof Botha joins SpaceX board

SpaceX has appointed former Sequoia Capital managing partner Roelof Botha to its board of directors, less than a week after the company's record-breaking IPO.

Anthropic brings Artifacts to Claude Code

The company has added Artifacts to Claude Code, allowing developers to turn coding sessions into interactive web pages that can be shared with their teams. The feature is available in beta for Claude Team and Enterprise customers.

Barret Zoph departs OpenAI after five months

Not long after co-founding Thinking Machines Lab, Barret Zoph left Mira Murati's startup to return to OpenAI. Now, just five months later, he is departing OpenAI as well, stepping down from his role leading the company's enterprise AI sales efforts.

OpenAI hires former Trump AI official Dean Ball

AI policy scholar Dean Ball is joining OpenAI to lead a newly formed Strategic Futures team focused on frontier AI policy and internal governance. Ball previously served as a senior AI policy adviser at the White House Office of Science and Technology Policy, where he helped shape early AI policies under the Trump administration.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

Father's Day is just a few days away, and if you're looking for a quick but thoughtful touch to add to dad's day, this might be it.

In today's issue, we'll show you how to create a custom Father's Day card using just a couple of photos and a single prompt.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• AI CEOs Meet G7 Leaders to Discuss Global AI Rules

• Midjourney Launches AI-Powered Full-Body Ultrasound Scanner

• AI Tutorial: Create a Beautiful Father's Day Card with ChatGPT

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

G7 AI SUMMIT

AI CEOs Meet G7 Leaders to Discuss Global AI Rules

CEOs of leading AI companies met with G7 leaders in France to discuss the future of artificial intelligence, international AI rules, and cooperation on the emerging technology.

Details:

• President Trump joined senior administration officials and G7 leaders for a closed-door AI working session that included executives from OpenAI, Anthropic, Google DeepMind, Meta, Mistral AI, Cohere, and Salesforce.

• The discussions focused on AI safety, cybersecurity risks, children's online safety, access to advanced AI models, and international coordination on AI governance.

• Anthropic CEO Dario Amodei and Google DeepMind CEO Demis Hassabis reportedly called for a U.S.-led coalition to coordinate global AI standards and manage the safe deployment of frontier AI systems.

• Sam Altman proposed an international forum that would develop common AI safety testing standards, provide independent assessments of model capabilities and risks, and facilitate cooperation among democratic nations.

• Trump described the meeting as "excellent," emphasizing both AI's enormous economic potential and the need to manage its risks while maintaining U.S. leadership over China.

Meanwhile, world leaders including French President Emmanuel Macron and Indian Prime Minister Narendra Modi voiced concerns over recent U.S. export restrictions on Anthropic's newest models, warning that the possibility of access to top American AI being cut off from one day to the next could harm both European economies and the AI firms themselves.

Source

TOGETHER WITH STORECLAW

Stop Paying for 6 Tools. One AI Does It All

Most e-commerce sellers are running their store across 6 to 8 separate tools — and paying hundreds of dollars a month for the privilege. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.

It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.

Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.

Free to start. No credit card required.

Try it for free today

MIDJOURNEY

Midjourney Launches AI-Powered Full-Body Ultrasound Scanner

Midjourney is expanding beyond AI image generation with Midjourney Medical, introducing an ultrasound-based full-body scanner designed to provide fast, radiation-free 3D body imaging and long-term health tracking.

Details:

• The Midjourney Scanner, the company's first hardware product, uses a ring of thousands of ultrasound transducers to capture detailed 3D images of the body's muscles, fat, bones, organs, and other internal structures in roughly 60 seconds.

• Developed in partnership with Butterfly Network, the system incorporates 40 Butterfly Ultrasound-on-Chip™ imaging modules and approximately two petaflops of computing power to reconstruct full-body scans from ultrasonic signals.

• Unlike CT scans or X-rays, the scanner relies on ultrasound, avoiding radiation exposure. Midjourney CEO David Holz said the long-term goal is to deliver image quality that is comparable to, or in some applications eventually exceeds, MRI.

• Midjourney envisions the scanner being used regularly, potentially as often as daily, to monitor changes in body composition, muscle growth, fat distribution, and other physiological metrics in response to diet, exercise, or lifestyle changes.

The company also plans to open a Midjourney Spa in San Francisco's Union Square before the end of 2027, featuring approximately 10 scanners alongside gyms, saunas, cold plunges, and hot tub-equipped scanning rooms to create a wellness-focused preventive health experience.

Source

AI Tutorial

Create a Beautiful Father's Day Card with ChatGPT

1. Go to ChatGPT and sign in.

2. Upload a photo of you and a photo of your dad.

3. Use this prompt:

Create a heartwarming Father's Day watercolor greeting card poster using the provided reference images. Feature the same real father and daughter from the photos, preserving their exact facial identity, expressions, and natural features without alteration. 

The father and the daughter are standing close together, sharing a soft, genuine smile. Emphasize a warm emotional bond—subtle affection, comfort, and pride. Transform the scene into a soft hand-painted watercolor illustration with delicate brushstrokes and a pastel color palette (blush pink, cream, peach tones). Add a scrapbook-style background with floral elements, doodled hearts, torn paper textures, and handwritten notes. Include elegant typography at the top: “Happy Father’s Day” Add small sentimental quotes around the design: “To the world you are a father, but to me you are my world.” “Thank you for your love, guidance, and endless support.” “My first hero, my forever inspiration.” Style: vintage greeting card, nostalgic, wholesome, soft lighting, subtle shadows, highly detailed watercolor faces, premium print-ready poster design, balanced composition, portrait orientation. 

Aspect Ratio: 9:16 

Quality Tags: ultra-detailed, high resolution, realistic watercolor, soft textures

You can alter the prompt to change the text or however you see fit.

AI Tools to check out

🖱️ Heyclicky: An AI buddy that lives on your Mac. Ask a question out loud and it walks you through whatever you're working on.

⚙️ Poetic: AI that learns your rules and follows them every time, handling end-to-end processes and expanding the breadth and depth of what you can automate.

👨‍💻 Softr: Build and launch custom AI-powered portals and internal tools in minutes, no code required.

✅ Freddy: An MCP server that plugs your wearables, CGMs, power meters, and gym apps straight into Claude, ChatGPT, and any AI agent that speaks MCP.

✨ Hera: AI motion designer that creates professional animations in seconds.

TOGETHER WITH STORECLAW

Stop Paying for 6 Tools. One AI Does It All.

Most e-commerce sellers juggle 6–8 tools and pay hundreds monthly to keep operations running. StoreClaw replaces the stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks profit 24/7. Connect your store and let AI handle the work — no prompts, no complex setup, no credit card required.

Try if for free today

AI Findings/Resources

🛰️ Here are 5 wild stats from SpaceX’s first week on the Nasdaq

💻 7 ways AI can help with your Linux system management

👀 "I'm the only nontechnical employee at an AI startup. There are still some tasks I won't trust AI to handle."

👉 The art of loop engineering

🤔 How AI companies can pay fair rates for the content they need

The latest in AI and Tech

Anthropic joins Frontier carbon removal coalition

Anthropic has become the first AI startup to join Frontier, a coalition that funds carbon removal projects, contributing to a new $915 million funding round that raises Frontier's total commitments to $1.8 billion.

OpenAI is retiring ChatGPT Pulse

The company announced it will discontinue Pulse, its personalized daily digest feature in ChatGPT, within the next 14 days. Users are encouraged to use the newly updated scheduled tasks feature to receive daily briefings instead.

Gemini co-lead Noam Shazeer leaves Google for OpenAI

Noam Shazeer, co-lead of Google's Gemini AI models, has announced he is joining OpenAI, just two years after Google paid $2.7 billion to poach him and part of his team from Character AI.

Apple to raise prices as AI boom drives up chip costs

Apple plans to raise product prices as soaring memory chip costs, driven by booming AI demand and supply constraints, make absorbing higher expenses unsustainable, outgoing CEO Tim Cook said. While Apple has not confirmed which products will be affected, analysts expect upcoming devices, including the iPhone 18, could see higher prices.

Google introduces gemini-powered home speaker with AI voice assistant

The company has announced the Google Home Speaker, its first standalone smart speaker since 2020 and the first device built around its Gemini AI voice assistant. The $99.99 speaker introduces more natural conversations, multi-step voice commands, and contextual follow-up interactions, while premium AI features are available through the new Google Home Premium subscription.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻

Welcome to another edition of Horizon AI,

In today's issue, we'll show you how to easily transform complex text and data-heavy documents into interactive visual mind maps using NotebookLM.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• Z. ai Launches Open-Source GLM-5.2 AI Model

• Pinterest Tests AI Shopping Assistant with New ‘Ask Pinterest’ App

• AI Tutorial: How to Create Mind Maps in NotebookLM

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

ZAI

Z. ai Launches Open-Source GLM-5.2 AI Model

Chinese AI startup Z.ai has released GLM-5.2, a new open-weights large language model designed for long-running software engineering and coding tasks, claiming performance that rivals or exceeds leading proprietary models while offering significantly lower deployment costs.

Details:

• Z.ai has released GLM-5.2, a 753-billion-parameter open-weights language model under the MIT license, allowing organizations to freely download, modify, and self-host the model.

• The model supports a 1 million-token context window, and users can choose between different "Thinking Modes," including a maximum-performance mode ("Max") and a more efficient "High" mode that delivers comparable results while using fewer output tokens.

• According to Z.ai, GLM-5.2 outperforms OpenAI's GPT-5.5 on several software engineering benchmarks, including SWE-bench Pro, FrontierSWE, MCP-Atlas, and other evaluations focused on long-horizon coding and autonomous tool use.

• The model is available through Hugging Face, Z.ai's API, and more than 20 developer tools and coding environments.

Z.ai also introduced developer subscription plans starting at $12.60 per month, with API pricing substantially lower than that of most leading AI providers.

Source

TOGETHER WITH STORECLAW

One AI. Every Tool Your Store Actually Needs.

Most e-commerce sellers are paying for 6 to 8 separate tools that don't talk to each other — and spending hundreds of dollars a month just to keep up. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.

It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.

Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.

Free to start. No credit card required.

Start for free today

PINTEREST

Pinterest Tests AI Shopping Assistant with New ‘Ask Pinterest’ App

Pinterest has launched a standalone experimental web app called Ask Pinterest that uses AI to help users discover products and ideas through conversational prompts instead of traditional keyword searches.

Details:

• The app is powered by Pinterest's proprietary Taste Graph, which maps users' interests, aesthetics, and preferences to deliver more personalized recommendations.

• Users can ask complex or multi-step questions (such as planning a dinner party or decorating a room over time) and receive tailored suggestions while maintaining context across conversations.

• When users sign in, the app can also incorporate their saved Pins and Boards to further personalize recommendations.

• Pinterest's new app was announced alongside marketer-focused updates, including a beta AI assistant in its U.S. Ads Manager, a new Performance+ Creative AI model, and an MCP infrastructure layer that will allow advertisers to manage campaigns through third-party agentic tools in a standardized way.

By launching Ask Pinterest as a separate application, the company can experiment with AI-powered shopping experiences without altering the core Pinterest app. Pinterest believes these results will help build future AI-powered experiences within its flagship app.

Source

AI Tutorial

How to Create Mind Maps in NotebookLM

1. Go to NotebookLM and open an existing notebook (or create a new one and upload your sources).

2. Select the Mind Map option and wait a few seconds.

3. In the Studio panel, you will find a new note with your generated mind map — click on it.

4. Use the built-in arrow buttons on the nodes to expand or collapse subtopics. Click the download button to save it as a PNG.

5. You can also click on any individual node within the map. The tool will instantly pull up the exact corresponding information from your original document to explain that specific topic in detail.

AI Tools to check out

📞 Bland: An enterprise voice AI platform that lets customers build, run, and monitor AI phone agents capable of holding real conversations at scale.

✅ Filently: AI document filing for Google Drive.

🌐 TasteLab: Extract any website's design DNA.

💪🏻 PeakRoutine: A proactive AI coach that learns your biology and build habits that actually work for you.

🤖 Uplift: AI agents for business teams, built and run for you.

TOGETHER WITH STORECLAW

Stop Paying for 6 Tools. One AI Does It All.

Most e-commerce sellers juggle 6–8 tools and pay hundreds monthly to keep operations running. StoreClaw replaces the stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks profit 24/7. Connect your store and let AI handle the work — no prompts, no complex setup, no credit card required.

Try if for free today

AI Findings/Resources

👀 OpenAI takes ChatGPT Images to the streets. Its new billboard campaign is now showing in Detroit, New York, Chicago, and Los Angeles.

🏡 How AI may be messing with home prices

💡 McKinsey: The seven operating truths of AI-native companies

🤔 AI doesn't feel. So why does it have something like emotions?

🎓 Google CEO Sundar Pichai gives Stanford's 2026 Commencement address, urging graduates to choose optimism, embrace difficult challenges, and pursue what genuinely excites them.

The latest in AI and Tech

Disney adopts Adobe Firefly AI for theme park design

Disney's Imagineering R&D team is using Adobe's Firefly Foundry platform to streamline the design and visualization process for its theme parks and experiences. The collaboration uses AI models trained on Disney's own creative assets to transform sketches into detailed concept art, convert 2D designs into 3D prototypes, and more.

Microsoft makes Copilot Cowork generally available worldwide

Developed in collaboration with Anthropic, the AI assistant is designed to handle complex workplace tasks such as analyzing large datasets, comparing documents, and supporting sales teams. The general release adds support for multiple AI models, including Anthropic's Opus 4.8, Sonnet 4.6, and GPT-5.5, with Microsoft's Cowork 1 model coming soon.

ChatGPT market share dips below 50% for the first time

OpenAI's ChatGPT remains the world's most widely used AI assistant, but its market share has fallen below 50% for the first time as competitors such as Google's Gemini and Anthropic's Claude continue to gain users. According to Sensor Tower, the AI assistant market is becoming more competitive, with users increasingly switching platforms based on features, integrations, and brand perception.

Trump to reject UK plea over Anthropic ban

The Trump administration is expected to reject the UK's request for an exemption from new export restrictions that block foreign access to Anthropic's most advanced AI models, citing national security concerns.

SpaceX nears $3 trillion valuation after Cursor deal

SpaceX's valuation briefly climbed to $2.9 trillion, surpassing Amazon before pulling back, following its planned $60 billion acquisition of Cursor and the start of options trading.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻