Forward Future by Matthew Berman

AI Hiring Drop, Claude Opus 4.7 Launch, & xAI Turns Cloud Provider

Matthew Berman — Fri, 17 Apr 2026 11:01:00 +0000

One Platform. Every Agent, App, and Workflow You’ll Ever Need

Good morning. It's Friday, April 17, and we're covering hiring’s 20% drop (and why AI isn’t to blame), Microsoft’s massive new AI datacenter push, Anthropic’s latest Claude 4.7 release, and more.

YOUR DAILY ROLLUP

One Platform. Every Agent, App, and Workflow You'll Ever Need.

Salesforce is leading the way as the trusted partner for the agentic era.

No-code and low-code tools let anyone build agents, prompts, and automations in minutes. Developers get a unified environment to build, customize, and test connected AI across a single platform. And it's fully open — connect your own data, bring your own models (OpenAI, Anthropic), and run across any cloud. One trusted foundation. Every workflow. Unprecedented speed to value.

See what's possible when your entire stack is unified. Explore Agentforce.

HIRING DECLINE

LinkedIn Data: Hiring Down 20% Since 2022, Not Due to AI

LinkedIn reports that global hiring has declined about 20% since 2022, according to chief legal officer Blake Lawit, speaking on April 15, 2026, at the Semafor World Economy summit. The company’s data—spanning over 1 billion users—shows no clear evidence that AI is driving the slowdown, despite widespread concern about job displacement.

Instead, Lawit pointed to higher interest rates as the more likely cause of reduced hiring. He added that entry-level hiring has not declined disproportionately compared to mid- or late-career roles. → Read the full article here.

BIOTECH

Science Corp. Nears First Human Brain Implant With Biohybrid Sensor

Max Hodak’s startup Science Corporation is preparing for its first human brain implant to test a new sensor, a foundational step toward its long-term goal of building a biohybrid brain-computer interface that blends lab-grown neurons with electronics. The company has enlisted Murat Günel to advise on U.S. clinical trials, which could begin as early as 2027. Unlike traditional implants, the initial sensor—containing 520 electrodes—will sit on the brain’s surface rather than penetrate tissue, potentially reducing damage. Science recently raised $230 million at a $1.5 billion valuation and is advancing both therapeutic devices (like its PRIMA vision implant) and longer-term human enhancement goals. → Read the full article here.

MODELS

Alibaba’s Qwen3.6–35B Model Open-Sourced With Sparse MoE Design

Alibaba’s Qwen team released Qwen3.6-35B-A3B on April 16, 2026, an open-source multimodal AI model using a sparse mixture-of-experts (MoE) architecture. The model has 35 billion total parameters but activates only 3 billion at runtime, aiming to deliver significantly higher efficiency without sacrificing performance. Qwen claims the model matches or exceeds larger systems on coding and vision-language benchmarks, including agentic coding tasks and spatial reasoning. → Read the full article here.

POWERED BY SALESFORCE

One Foundation for Every Agent, Model, and Workflow.

The Salesforce Platform lets you build AI agents, connect your own data, and run across every cloud – all from one unified environment. Open, extensible, and enterprise-ready. See what’s possible when your entire stack is unified.

Explore Agentforce.

NEWS

What Else is Happening

Cursor Adds Claude Opus 4.7: Cursor launches Anthropic’s latest model with improved autonomy and creative reasoning, offering a limited-time 50% discount to accelerate developer adoption and experimentation.

OpenAI Revamps Codex: The revamp gives Codex the ability to run background agents on a user's desktop, opening apps and operating a cursor without disrupting active work.

ASML, TSMC Boost AI Outlook: Strong forecasts from ASML and TSMC signal sustained AI chip demand, with cloud giants set to exceed $600 billion in 2026 data center spending despite ongoing capacity constraints.

Runway Launches Seedance 2.0 API: Runway releases Seedance 2.0 via its API, expanding developer access to its generative video tools and enabling broader integration into production and creative workflows.

Windsurf Unveils Agent Command Platform: Windsurf 2.0 introduces a unified command center, project “Spaces,” and cloud-based Devin agents, enabling developers to delegate tasks and maintain workflows asynchronously.

Ulysses Raises $46M for Ocean Tech: Ulysses secures $46 million led by Andreessen Horowitz to build technology for ocean exploration, infrastructure, and defense, targeting a largely uncharted domain critical to global systems.

FRIDAY FACTS

An Alcohol-Rich, Raspberry-Scented Space Cloud

Deep in space, molecular clouds contain massive volumes of alcohol, such as a cloud in the constellation Aquila packed with ethanol. Another cloud 26,000 light-years away, Sagittarius B2, contains ethyl formate, which gives raspberries their smell and rum its taste.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Jagged AI, Compute Economy Shift, & Gemini Desktop Push

Matthew Berman — Thu, 16 Apr 2026 11:01:00 +0000

Good morning. It's Thursday, April 16, and we're covering AI’s uneven intelligence, the rise of a compute-powered economy, Google bringing AI directly into the desktop experience, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

Matt sits down with Aravind Srinivas, CEO of Perplexity AI

MARKET PULSE

Allbirds Sells Shoe Brand, Rebrands As AI Cloud Provider NewBird

After selling its footwear brand and assets to American Exchange Group for $39 million in March 2026, Allbirds is rebranding as NewBird AI and pivoting to AI infrastructure. The company announced a $50 million convertible financing facility from an undisclosed institutional investor and plans to offer GPU-as-a-service and AI-native cloud solutions.

The move allows the Nasdaq-listed shell (ticker: BIRD) to redirect into the high-demand AI compute market. The transition remains subject to shareholder approval, with a vote scheduled for May 18, 2026, and a potential dividend payout expected in Q3 if approved. → Continue reading here.

INTELLIGENCE

“Jagged Intelligence” Reframes AI As Uneven, Not Humanlike

A growing group of researchers and economists argue that artificial intelligence is best understood as “jagged intelligence,” a term coined by Andrej Karpathy to describe systems that excel in some tasks while failing at others. Recent examples highlight the contrast: AI models solved five of six problems at the 2025 International Math Olympiad but struggled with simple real-world reasoning questions.

The concept is reshaping how experts evaluate AI’s impact on jobs, suggesting automation will target specific tasks rather than entire professions. While capabilities are improving rapidly, uneven performance makes it difficult to predict where systems will succeed or break down. → Read the full article here. (Paywall)

COMPUTE

Greg Brockman Outlines Shift to AI-Driven, Compute-Powered Economy

Greg Brockman says the global economy is shifting toward a “compute-powered” model, where access to AI and processing power increasingly determines productivity. He argues that AI is already accelerating software engineering and is poised to transform all computer-based work by reducing the need for step-by-step human instruction.

Tools like ChatGPT and Codex are enabling individuals and small teams to build faster and tackle more complex projects, lowering barriers between ideas and execution. Brockman notes this shift could disrupt job stability and institutions, while also expanding entrepreneurship and creative output. He emphasizes that how these systems are deployed—and who benefits—remains an open question. → Read the full article here.

NEWS

What Else is Happening

Snap Cuts 1,000 Jobs Citing AI: Snap Inc will cut 16% of its workforce after pressure from activist investor Irenic Capital and a 30% stock drop, citing AI-driven efficiency gains and halted hiring.

Claude Adds Multi-Session Workspace: Redesigned Claude Code brings side-by-side sessions, integrated terminal, and previews, aiming to streamline developer workflows with customizable layouts and native plugin support.

Emergent Launches Wingman AI Agent: India’s Emergent unveils a messaging-based autonomous agent that executes tasks across apps via chat, targeting workflow automation as competition intensifies in the AI agents market.

Duolingo Drops AI Usage Metrics: Duolingo will no longer evaluate employees on AI usage after backlash and confusion, shifting performance reviews to outcomes as CEO Luis von Ahn admits overemphasis on AI adoption.

Adobe Unveils Firefly AI Assistant: Adobe launches an agent that runs Photoshop, Premiere, and more from one prompt, coordinating 100+ tools to automate workflows as it pushes deeper into AI-driven creative software.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Claude Automation, AI Divide, & NVIDIA's Quantum Push

Matthew Berman — Wed, 15 Apr 2026 11:01:00 +0000

Good morning. It's Wednesday, April 15, and we're covering Claude’s push into automated coding workflows, a growing AI divide between experts and the public, NVIDIA’s latest quantum leap, and more.

YOUR DAILY ROLLUP

Your Personal Context Is the Next AI Race

As AI levels the playing field, advantage shifts from model access to what those models are grounded in. Your knowledge.

Recall 2.0 is one place for your knowledge and AI. Save what matters, write what you think, curate an AI that knows what you know.

"Condense my research on LLMs, enrich it with new studies, find the exact point quantization was mentioned in my podcasts."

Talk to your knowledge, the internet, or both in the model you choose.

The future belongs to AI that works from what you know.

Try Recall 2.0 free today or use code FF25 for 25% off the uncapped plan.

REAL OR AI

Think You Can Still Spot AI? Answer ↓

PRODUCT RELEASE

Anthropic Introduces Claude Code “Routines” for Automated Workflows

Anthropic has launched “routines” for Claude Code, a feature that lets users automate development tasks using scheduled jobs, API calls, or GitHub events. Each routine bundles a prompt, repositories, and connectors, then runs autonomously in cloud-hosted sessions—even when a user’s machine is offline.

The system supports triggers like cron-style schedules, HTTP endpoints, and repository activity such as pull requests or pushes. The feature turns Claude into a persistent automation layer for engineering workflows, from code review to incident response. The capability is currently in research preview, with evolving limits and API behavior. → Read the full article here.

REPORT

Stanford Report Finds Widening Gap Between AI Experts and Public

A new Stanford University AI report released April 13, 2026, finds a growing divide between how experts and the public view artificial intelligence, particularly in the U.S. While a majority of AI experts expect positive long-term impacts across areas like healthcare, jobs, and the economy, public sentiment is far more skeptical.

Only 10% of Americans say they are more excited than concerned about AI, with many worried about job loss, rising energy costs, and broader economic disruption. The report also highlights declining optimism among younger users, even as AI adoption remains high. → Read the full article here.

CHIPS

NVIDIA Says AI Turns Months-Long Chip Design Work Into Overnight Task

NVIDIA says it is using artificial intelligence to dramatically accelerate chip design, cutting a task that once took eight engineers 10 months down to a single overnight run on one GPU. The company’s chief scientist, William Dally, described how AI systems—particularly reinforcement learning models like NB-Cell—now handle labor-intensive steps such as standard cell library design.

NVIDIA is also deploying internal large language models trained on proprietary GPU data to assist engineers and boost productivity. Despite these gains, the company says fully autonomous chip design remains far off, with human oversight still required for verification and system-level decisions. → Read the full article here.

ROBOTICS

Google Unveils Gemini Robotics-ER 1.6 To Boost Real-World Robot Reasoning

Google has introduced Gemini Robotics-ER 1.6, an upgraded AI model designed to improve how robots understand and act in physical environments. The system enhances “embodied reasoning,” combining spatial awareness, multi-view perception, and task planning to enable more autonomous robotic behavior. New capabilities include instrument reading—allowing robots to interpret gauges and sensors—and improved success detection to determine when tasks are complete. → Read the full article here.

NEWS

What Else is Happening

Texas man charged after Altman attack: Daniel Moreno-Gama allegedly threw a Molotov cocktail at Sam Altman’s home, citing AI extinction fears; prosecutors filed attempted murder and federal explosives charges.

OpenAI acquires finance startup Hiro: OpenAI bought AI financial planning startup Hiro, founded in 2024 by Ethan Bloch, likely as a ~10-person acqui-hire; the service shuts down April 20 with data deleted May 13.

AI designs ultra-strong rustproof steel: Researchers used machine learning to create a 3D-printable alloy reaching 1,713 MPa strength and 15% ductility, outperforming stainless steel and improving manufacturing efficiency.

Google Chrome adds reusable AI Skills: Google launched “Skills” in Chrome, letting users save and reuse Gemini prompts across sites; rolling out April 14, 2026 to desktop users with English (US) settings.

US Treasury probes Anthropic’s Mythos AI (Paywall): Treasury officials, led by CIO Sam Corcos, seek access to Anthropic’s restricted model to identify security flaws as agencies prepare for risks from advanced AI systems.

REAL OR AI

The Image Is AI. Tune in next week for another Real or AI test.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

US-China AI Race, Worker Divide, & OpenAI UK Expansion

Matthew Berman — Tue, 14 Apr 2026 11:01:00 +0000

The Box Agent: unleash your enterprise content with AI

Good morning. It's Tuesday, April 14, and we're covering China’s rise, workplace AI adoption, OpenAI doubling down on its London expansion, and more.

YOUR DAILY ROLLUP

The Box Agent - Turn Your Content Into Action

New from Box: The Box Agent. A fundamental shift in how enterprises use content to autonomously execute complex tasks. Using the latest advanced reasoning models, the Box Agent helps organizations put unstructured data to work to unlock critical insights from their content, automate tasks, accelerate decisions, and help teams work smarter — while keeping sensitive content protected every step of the way.

Your business lives in your content. Learn how the Box Agent unleashes it.

VIDEO

Cut AI Costs With Local Models

GEOPOLITICS

U.S. Can’t Halt China’s AI Rise

In an April 13, 2026 opinion essay, Sebastian Mallaby argues that U.S. export controls on advanced semiconductors have failed to slow China’s artificial intelligence progress. Chinese firms have bypassed restrictions by renting overseas computing power, stacking lower-end chips, and rapidly copying leading models through “distillation.” Mallaby contends the U.S. and China are now roughly level in AI capabilities, with China potentially ahead in industrial deployment.

He proposes shifting strategy toward a global AI safety agreement—modeled on nuclear nonproliferation—to manage shared risks rather than attempting to contain China’s advances. → Read the full article here. (Paywall)

WORKFORCE

AI Adoption Splits U.S. Workers As Productivity Rises, Fears Grow

A Gallup poll finds growing but uneven adoption of artificial intelligence in U.S. workplaces, with about 3 in 10 employees using it frequently and roughly half rarely or never using it. Among those with access, two-thirds report productivity gains, especially in management, healthcare, and tech roles.

However, skepticism persists: many workers cite ethical concerns, data privacy risks, or simply prefer existing workflows. Concern about job loss is also rising, with 18% of workers saying automation could eliminate their job within five years, up from 15% in 2025. → Read the full article here.

ENERGY

AI Boom Spurs Renewed Push for Carbon Capture Projects

Rising electricity demand from AI data centers is driving new interest in carbon capture technology for natural gas power plants, with at least five U.S. projects under consideration involving companies like Google, ExxonMobil, Chevron, and Meta. Long viewed as too expensive, carbon capture is gaining momentum as tech giants seek to meet climate goals while securing reliable energy.

Executives say deep-pocketed “hyperscalers” could help scale the technology, though no U.S. gas plants currently use it. Analysts estimate up to $80 billion could be invested if projects move forward, but high upfront costs and reliance on federal tax credits remain major hurdles. → Read the full article here.

NEWS

What Else is Happening

Claude Mythos Alarms UK Regulators: Anthropic claims its unreleased model can exploit zero-day flaws across major systems; UK authorities plan urgent talks as experts debate whether risks are overstated or credible.

NYT Reshuffles Leadership Around AI Push: The New York Times promotes Alex Hardiman and Hannah Yang to executive VPs, expands AI and data oversight, and begins a CTO search as it doubles down on subscriptions.

Harvey Launches End-to-End Legal AI Agents: Startup unveils agents that plan, research, and draft legal work across practice areas, producing cited memos and reports while lawyers retain final judgment and oversight.

Vercel Signals IPO as AI Drives Growth: CEO Guillermo Rauch says the company is “ready” to go public as ARR jumps to $340 million, fueled by AI-generated apps, with agents now creating 30% of deployments.

Meta Set to Overtake Google Ads: Emarketer projects Meta will hit $243.46 billion in 2026 ad revenue, surpassing Google’s $239.54 billion, driven by Advantage+ tools and faster growth rates.

PROMPT OF THE WEEK

The "What am I missing?" Prompt

Ask it at the end of any plan, email, strategy, or decision. It consistently surfaces blind spots you didn't think to ask about directly.

A few others in the same spirit:

Instant steelman: "What would a skeptic say about this?"
Force direct answers: "What's the most likely way this goes wrong?"
Ask for summaries: "Explain what's actually happening here."

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

OpenAI's “Startup-in-a-Box,” Agent Risks, & Benchmark Exploits

Matthew Berman — Mon, 13 Apr 2026 11:01:00 +0000

Good morning. It's Monday, April 13, and we're covering the risks of autonomous agents, how benchmark loopholes are skewing AI performance, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

THE WEEKEND RECAP

Matt Sits Down With Tech Legend Marc Benioff

THE FUTURE LIVE

Can Meta Win the AGI Race?

BEP Holdings founder argues Zuck's path looks different: free devices, localized AI for underserved markets, and a data moat nobody else has.

AGENTS

AI Agents Like Claude and OpenClaw Raise Power—and Risk

A new wave of autonomous AI agents—including Anthropic’s Claude Cowork and open-source tool OpenClaw—is accelerating the shift from chatbots to systems that can act independently across software environments. These agents can perform complex tasks like coding, legal document review, and inbox management with minimal human input, but require deep system access to do so.

That autonomy introduces new risks, from faulty decisions to data exposure, especially in open-source systems without centralized oversight. The result is a growing tension between capability and control, as enterprises weigh productivity gains against reliability and security concerns. → Read the full article here.

VENTURES

OpenAI Pitches With AI Tools and Funding

OpenAI outlined a policy proposal to support entrepreneurship through a “startup-in-a-box” model—an AI-powered bundle of tools, infrastructure, and potential funding. The concept includes automated back-office services like accounting, marketing, and procurement, along with model legal contracts, aiming to reduce barriers for new founders.

OpenAI suggests pairing these tools with microgrants or revenue-based financing to accelerate company creation. The proposal comes as competition intensifies, with founders increasingly turning to rival systems like Anthropic’s Claude. It also calls for clearer government rules on AI use, particularly around safety and ensuring public access to foundational models. → Read the full article here.

SPENDING

SpaceX Posts $5B Loss As AI Spending Surges

SpaceX reported nearly $5 billion in losses for 2025 despite generating over $18.5 billion in revenue, driven largely by heavy investment in AI through its newly acquired xAI unit. The company spent approximately $13 billion on chips and data centers for AI—about 50% more than its combined rocket and satellite divisions.

While its core businesses, including Starlink and launch services, produced roughly $8 billion in earnings, those gains were offset by depreciation and AI-related capital expenditures. The figures underscore how SpaceX’s upcoming IPO will effectively ask investors to fund Elon Musk’s AI ambitions alongside its dominant space business. → Read the full article here. (Paywall)

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

AI Agents Rise, Model Efficiency Breakthrough, & Job Shifts

Matthew Berman — Fri, 10 Apr 2026 11:01:00 +0000

Let Accio Work — Try One Week Free!

Good morning. It's Friday, April 10, and we're covering AI agents replacing software interfaces, breakthrough model efficiency, how AI is reshaping jobs, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

The Pentagon Once Built a Powerful Supercomputer for $2 Million

A comparable system built the normal way would have cost ~$20 million or more. The hardware they used? You almost certainly own one. Answer ↓

POWERED BY ACCIO WORK

Accio Work: Your Agentic Team for Real Business

Meet Accio Work - the agentic workspace for business owners and solopreneurs. Our specialized agent team manages operations for you—sourcing products, negotiating with suppliers, managing stores, and launching marketing—all on autopilot.

With verified skills and business tool APIs, Accio Work takes action while you stay in control. Powered by Alibaba.com’s 1B+ products and global supplier data, it handles product selection and execution seamlessly. No setup required—just results.

Try Accio Work now — 7 Days Free!

VIDEO

Anthropic Banned OpenClaw…

Anthropic bans third-party Claude tools amid GPU crunch, sparking backlash, confusion, and rapid shift to GPT alternatives.

THE FUTURE LIVE

INTERFACES

Sierra CEO Bret Taylor Bets on AI Replacing Click-Based Software

At the HumanX conference, Bret Taylor, CEO of Sierra, said traditional click-based software interfaces could soon give way to natural language interactions. The company recently launched Ghostwriter, an “agent as a service” platform that lets users describe tasks and automatically generate AI agents to execute them.

Taylor argues this shift could eliminate the need to navigate complex enterprise tools like Workday. Sierra claims rapid adoption, including deploying an AI agent for Nordstrom in four weeks, and reaching a $100M annual revenue run rate within 21 months. Still, industry experts note current AI agents often require ongoing human tuning, limiting full autonomy. → Read the full article here.

HARDWARE

Ex-Apple Engineers Debut AI “Button” Device, Face Purpose Questions

Two former Apple engineers unveiled “Button,” a small AI wearable resembling an iPod Shuffle that activates a chatbot with a physical press. The device, created by ex-Apple Vision Pro team members, aims to offer a privacy-conscious alternative to always-on AI wearables by requiring user input to listen. It can respond via speaker or connect to earbuds and smart glasses over Bluetooth.

However, its creators have not clearly explained why the product should exist as dedicated hardware rather than a smartphone app. The launch follows poorly received AI devices like the Humane AI Pin and Rabbit R1, raising skepticism about the category. → Read the full article here.

NEWS

What Else is Happening

Intel, Google Expand AI CPU Push: The companies deepened ties to deploy Xeon 6 chips and co-develop infrastructure processors, targeting rising demand for AI inference and multi-step agentic systems.

CIA Reportedly Uses “Ghost Murmur” AI: Lockheed Martin system allegedly detected heartbeats across miles to locate a downed U.S. airman in Iran, highlighting emerging AI surveillance capabilities in military operations.

Florida Probes OpenAI Before IPO: Florida’s attorney general launched an investigation into OpenAI over security and misuse concerns, as the company eyes a potential $1 trillion public offering.

CoreWeave Lands $21B Meta Deal: CoreWeave signed a cloud agreement through December 2032 to supply AI compute capacity, underscoring surging infrastructure demand as Meta scales advanced AI workloads.

FRIDAY FACTS

It Was 1,760 PlayStation 3s

In 2010, the U.S. Air Force Research Laboratory wired together 1,760 Sony PS3 gaming consoles in Rome, New York and called it the Condor Cluster. At its peak, it ranked 33rd on the list of the world's most powerful supercomputers.

It could process 500 trillion floating-point operations per second. The Air Force Research Laboratory utilized the cluster for high-resolution satellite imagery analysis, pattern recognition, and early AI research.

Each PS3 ran about $400. Equivalent computing hardware cost $10,000 per unit. The full build came in at $2 million — roughly 10% of what a conventional system would have cost — and consumed 10% of the energy.

The PS3's IBM Cell processor was the key. It was built to handle massive amounts of simple math simultaneously, which made it near-perfect for parallel computing at scale.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Anthropic’s Mythos, Managed Agents, & Meta’s Muse AI Launch

Matthew Berman — Thu, 09 Apr 2026 11:01:00 +0000

Good morning. It's Thursday, April 9, and we're covering Anthropic’s locked-down Mythos model, its new agent architecture, Meta’s latest push into multimodal AI, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

Recraft V4: Design at Full Precision

Recraft V4 introduces two powerful model families — Raster for high-resolution photorealistic scenes or detailed illustrations and Vector for scalable SVG graphics with design taste at the core of every output. Available inside Recraft Studio, Recraft V4 supports complex prompts, crisp typography, and professional-grade outputs across branding, packaging, and digital workflows.

Start creating with Recraft V4.

VIDEO

Mythos Stuns with Superhuman Coding

CYBERSECURITY

Anthropic Withholds Claude Mythos Over Unprecedented Hacking Risks

Anthropic has developed a new AI model, Claude Mythos, capable of uncovering thousands of previously undetected software vulnerabilities—including flaws dating back 27 years—and is choosing not to release it publicly. Announced on April 7, 2026, the model’s capabilities raised concerns that it could accelerate cyberattacks if misused.

Instead, Anthropic is sharing Mythos with select partners like CrowdStrike, Palo Alto Networks, Amazon, Apple, and Microsoft under a defensive initiative called “Glasswing.” The move highlights a growing tension in AI: tools that strengthen cybersecurity can also empower attackers at unprecedented speed and scale. → Read the full article here.

AGENTS

Anthropic Details “Managed Agents” System to Future-Proof AI Workflows

Anthropic describes a new architecture for long-running AI systems called “Managed Agents,” designed to separate core components—model, execution environment, and memory—to improve reliability and adaptability. The system decouples the “brain” (Claude model and harness), “hands” (tools and sandboxes), and “session” (event log), allowing each to scale or fail independently.

This shift reduced latency significantly, with time-to-first-token dropping ~60% at median levels and over 90% at the high end. The approach also strengthens security by isolating credentials from AI-generated code and supports more flexible deployments, including enterprise environments. → Read the full paper here.

EFFICIENCY

Heat-Resistant Memristor Chip Operates at 700°C, Boosts AI Efficiency

Engineers at the University of Southern California have developed a memristor-based memory device that operates at temperatures up to 700°C (1,300°F), far beyond the ~200°C limit of conventional electronics. Published on March 26, 2026 in Science, the device uses a tungsten–hafnium oxide–graphene structure to prevent heat-induced failure and maintain performance.

It can store data for over 50 hours, endure more than one billion cycles, and compute at low voltage with nanosecond speeds. The breakthrough could enable AI processing in extreme environments like space, geothermal systems, and nuclear settings, while significantly improving energy efficiency for matrix-heavy workloads. → Read the full article here.

3D WORLDS

World Labs Launches Marble 1.1 and 1.1-Plus for Scalable 3D Worlds

World Labs announced two updates to its Marble model family, aimed at improving visual quality and scaling capabilities for AI-generated environments. Marble 1.1 enhances lighting, contrast, and reduces visual artifacts, while Marble 1.1-Plus is designed to generate larger, more complex virtual worlds.

The models are available via a user interface and API, signaling a push toward both creator-friendly tools and developer integration. The release reflects growing competition in generative 3D and simulation platforms. → Read the full article here.

NEWS

What Else is Happening

OpenAI Unveils Child Safety Blueprint: OpenAI introduced a U.S.-focused framework targeting AI-enabled child exploitation, emphasizing updated laws and built-in safeguards developed with NCMEC and safety groups.

Apple Quietly Opens Door to NVIDIA: Apple approved a third-party driver enabling NVIDIA GPUs on Apple Silicon Macs in limited external setups, signaling cautious flexibility as AI workloads increasingly demand NVIDIA.

Bezos AI Lab Poaches Key Talent (Paywall): Jeff Bezos’s Project Prometheus hired Kyle Kosic from OpenAI to build industrial-focused AI systems, intensifying competition for talent and expanding into real-world intelligence.

Google Finance Expands with AI Globally: Google is rolling out its AI-powered Finance platform to 100+ countries, adding natural-language research, advanced charts, real-time data, and live insights with multilingual support.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Meta's Token Wars, AI Gig Jobs, & Musk’s Chip Push

Matthew Berman — Wed, 08 Apr 2026 11:01:00 +0000

Let Accio Work — Try One Week Free!

Good morning. It's Wednesday, April 8, and we're covering Meta’s AI token race, older workers pivoting to AI gigs, Intel teaming up with Elon Musk, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

Accio Work: Your Agentic Team for Real Business

Try Accio Work now — 7 Days Free!

REAL OR AI

Think You Can Still Spot AI? ↓ Real or fake?

VIDEO

Matt Builds Something… Journey!

PRODUCTIVITY

Meta Employees Compete for AI “Token Legend” Status Internally

Meta employees are competing on an internal leaderboard that ranks workers by how many AI tokens they consume, with top users earning titles like “Token Legend” and “Session Immortal.” The system, dubbed “Claudeonomics,” tracked more than 60 trillion tokens used across roughly 85,000 employees in a recent 30-day period. Leadership has encouraged heavy AI usage as a proxy for productivity, with some executives suggesting engineers should spend hundreds of thousands of dollars annually on tokens.

The trend reflects a broader Silicon Valley shift toward “tokenmaxxing,” where AI usage is gamified and treated as a performance signal. Critics and uncertainties remain around actual cost efficiency and whether higher token usage directly correlates with better outcomes. → Read the full article here. (Paywall)

WORKFORCE

Older Workers Turn to AI Training As Fallback Employment

A growing number of older, skilled U.S. workers are turning to AI data training jobs after struggling to re-enter the workforce following layoffs or health setbacks. The work—labeling and evaluating outputs for models like ChatGPT and Google Gemini—offers flexible, contract-based income, but often at significantly lower pay and without benefits.

Workers over 60 take about 50% longer to find new jobs, and many never regain prior earnings, making AI training a “bridge job” to stay afloat. While some experts can earn over $100 per hour, many roles pay $20–$40 hourly and lack stability. → Read the full article here.

ROBOTICS

Generalist Unveils GEN-1 Robot Model With 99% Task Reliability

Robotics startup Generalist has introduced GEN-1, a physical AI model that achieves 99% success rates on repetitive, fine-motor tasks like folding boxes, packing devices, and repairing robot vacuums. The system is trained on over 500,000 hours of human “data hands” interactions—capturing detailed motion and visual input—and can adapt to new tasks with about an hour of additional robot-specific training.

Unlike earlier systems, GEN-1 can recover from disruptions and improvise actions not explicitly programmed, such as adjusting grip or retrying failed steps. → Read the full article here.

NEWS

What Else is Happening

Dimon Predicts 3.5-Day Workweek: JPMorgan CEO Jamie Dimon says AI could shrink workweeks within 30 years while boosting jobs, but warns rapid automation may cut roles and demands retraining efforts

Anthropic Secures Massive TPU Expansion: Anthropic partners with Google and Broadcom for multi-gigawatt TPU capacity starting 2027, fueling Claude amid $30 billion revenue surge and rapidly growing enterprise demand.

Samsung Profit Surges on AI Chips: Samsung forecasts 57.2 trillion won Q1 operating profit, up eightfold year-over-year, as booming AI memory demand drives shortages, higher prices, and renewed competitiveness in memory.

Rocket Launches AI Strategy Reports: India’s Rocket unveils a platform generating consulting-style product strategies from prompts, offering pricing, insights, and competitor tracking at far lower cost than traditional firms.

REAL OR AI

The Image Is Real. Tune in next week for another Real or AI test.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Altman’s AI Plan, OpenAI in Turmoil, & Meta’s Model Shift

Matthew Berman — Tue, 07 Apr 2026 11:01:00 +0000

Good morning. It's Tuesday, April 7, and we're covering Altman’s policy push on AI, internal shakeups at OpenAI, Meta’s hybrid model strategy, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

I Was Hacked…

AI hacker Ply tries to break into Matt’s personal AI system—testing jailbreaks, token attacks, and security defenses.

THE FUTURE LIVE

Founder of BEP Research Joins The Live Show

POLICY

Sam Altman Proposes “AI New Deal” With Taxes, Public Fund

OpenAI CEO Sam Altman has outlined a 13-page policy blueprint urging governments to prepare for rapid advances in AI, including potential “superintelligence,” according to an Axios interview. The proposal calls for measures like a public wealth fund, robot taxes, and a four-day workweek to offset economic disruption, including job losses and shifts in tax revenue.

Altman warns that near-term risks include major cyberattacks and the possible misuse of AI in biological threats. The plan positions OpenAI as both a builder of transformative technology and a participant in shaping its regulation—highlighting urgency while inviting broader debate. → Read the full article here.

LEADERSHIP

Inside OpenAI Turmoil: Allegations Challenge Sam Altman’s Leadership

A New Yorker investigation published April 6, 2026 details internal conflicts at OpenAI, including allegations that CEO Sam Altman misled colleagues and obscured safety concerns as the company advanced powerful AI systems. Internal memos from chief scientist Ilya Sutskever and others accused Altman of a “pattern” of deceptive behavior, contributing to his brief ouster in November 2023 before he was reinstated within days.

The investigation exposes tensions between rapid commercialization and AI safety, with some researchers warning that safeguards were deprioritized. The episode underscores broader questions about governance, accountability, and trust in companies building potentially transformative AI technologies. → Read the full article here.

CYBERSECURITY

AI Arms Race Reshapes Cybersecurity as Attacks Grow Autonomous

Cybersecurity experts warn that new AI systems from companies like Anthropic and OpenAI are accelerating both cyberattacks and defenses, according to a New York Times report published April 6, 2026. Anthropic disclosed a case in which suspected Chinese state-backed hackers used AI agents to automate up to 80–90% of an intrusion campaign across roughly 30 organizations.

While such incidents remain rare, upcoming AI releases are expected to make it far easier to discover and exploit vulnerabilities at scale. At the same time, defenders are deploying AI to detect weaknesses faster, setting up an escalating “AI vs. AI” dynamic. The core question: whether attackers or defenders gain the upper hand first. → Read the full article here. (Paywall)

NEWS

What Else is Happening

AI Fuels Cyber Arms Race (Paywall): A New York Times report finds AI automates up to 90% of some intrusions while also uncovering hundreds of zero-day flaws faster, intensifying attacker-defender competition.

Celigo Unveils AI Agent Platform: Celigo launches Ora and a low-code Agent Builder, letting non-technical teams automate workflows via natural language with governance controls, addressing struggles to scale AI beyond pilots.

Xoople Raises $130M for Earth AI Data: Spain’s Xoople secures $130 million Series B to build a satellite network delivering higher-precision geospatial data for AI, targeting enterprise demand for reliable “ground truth.”

Madison Air Targets $13.2B IPO: Ventilation firm Madison Air plans to raise up to $2.23 billion in a U.S. IPO, fueled by AI-driven data center cooling demand despite volatile markets slowing listings.

White House Sees AI Lowering Rates: Adviser Kevin Hassett said April 6, 2026 that AI-driven productivity and capital spending could ease inflation, giving the Federal Reserve room to cut interest rates.

PROMPT OF THE WEEK

Data Analysis Brief

Before analyzing this data, work through these steps in order:

1. ASSUMPTIONS — State what you're assuming about the data (collection method, time period, unit of measure, population it represents). Flag anything ambiguous.

2. QUALITY CHECK — Identify missing values, outliers, potential sampling bias, or anything that would limit the reliability of conclusions. Be specific.

3. FINDINGS — Separate into two buckets:
   - Descriptive: what the data directly shows (with numbers)
   - Interpretive: what you believe it means and why

4. LIMITS — What questions does this data explicitly cannot answer? What would you need to answer them?

5. HYPOTHESES — Generate 3 follow-up hypotheses worth testing, ranked by potential impact.

Do not skip steps or merge them. If you don't have enough information for a step, say so explicitly rather than proceeding.

[paste your data or describe your dataset here]

TWEETS

Artemis II Lunar Flyby

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

DeepSeek on Huawei, UK AI Shakeup, & Perplexity's Tax Tools

Matthew Berman — Mon, 06 Apr 2026 11:01:00 +0000

Good morning. It's Monday, April 6, and we're covering China’s push for homegrown AI chips, a UK shakeup at its top AI institute, the rise of AI-powered tax assistants, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

THE WEEKEND RECAP

Recraft V4: Design at Full Precision

Start creating with Recraft V4.

VIDEO

Google Just Dropped Gemma 4

MARKET PULSE

OpenAI and Anthropic Eye IPOs To Fund Trillion-Dollar Scale

Both OpenAI and Anthropic are exploring public listings despite access to abundant private capital, according to reporting by Axios. The driver isn’t liquidity—private markets and sovereign wealth funds already provide that—but the unprecedented scale of their ambitions, with each targeting trillion-dollar valuations and rapidly doubling revenue and costs.

Training frontier AI models requires massive, sustained spending on chips, talent, and infrastructure that could outstrip private funding depth. Going public would open deeper capital pools and create more flexible acquisition currency. → Continue reading here.

GEOPOLITICS

DeepSeek V4 To Run on Huawei Chips Amid China AI Push

Chinese AI startup DeepSeek is preparing to launch its V4 model in April 2026, designed to run on chips from Huawei—a milestone in China’s effort to reduce reliance on U.S. semiconductor technology. Major firms including Alibaba, ByteDance, and Tencent have reportedly ordered hundreds of thousands of Huawei’s new Ascend 950PR chips, driving prices up 20%.

The shift comes as U.S. export controls limit access to advanced chips from NVIDIA, forcing Chinese companies to adapt models for domestic hardware. Rewriting software to run efficiently on Huawei chips has delayed the launch but could give China tighter control over its AI stack. This move marks China’s shift toward a fully domestic AI stack: chips, models, and infrastructure. → Read the full article here. (Paywall)

ROBOTICS

Japan's Labor Crisis Has a Robot Solution

Japan isn't adopting physical AI because it wants to — it's adopting it because it has no choice. With its working-age population projected to fall by nearly 15 million over the next two decades, companies are deploying AI-powered robots across factories, warehouses, and critical infrastructure just to keep operations running. The government has backed the push with $6.3 billion in commitments and a target to capture 30% of the global physical AI market by 2040. This isn't a productivity play. It's a continuity tool.

What makes Japan's position interesting is the structural advantage it brings to a technology still being defined. The country has long dominated core robotics components — actuators, sensors, control systems — even as the U.S. and China race to build full-stack platforms. The emerging ecosystem looks less winner-take-all and more symbiotic: large incumbents like Toyota and Mitsubishi handling scale, while startups drive innovation in orchestration software and system design. The country that perfected the art of making things may be best positioned to make AI physical.→ Read the full article here.

OVERSIGHT

UK Orders Overhaul of Alan Turing Institute Strategy and Governance

The Alan Turing Institute has been told to make “significant” changes after a review by UK Research and Innovation found shortcomings in strategy and value for money. The institute, which received a £100 million, five-year funding package in 2024, faces pressure to better align with national priorities, particularly defense and security.

The warning follows governance concerns raised by a whistleblower complaint and leadership turnover, including the departures of CEO Jean Innes and chair Doug Gurr. While the review acknowledged strong scientific output, it concluded the institute lacks clear strategic focus and effective delivery. The government is now pushing for tighter oversight and a sharper mission tied to national resilience. → Read the full article here.

EMPLOYMENT

Economists Shift on AI Job Risk, Warn Policymakers Unprepared

Economists are increasingly taking AI’s potential impact on jobs seriously, even as current labor market data shows little disruption, according to reporting from The NY Times. Recent advances—such as reasoning-capable models and autonomous AI agents—have shifted views, raising concerns about future job losses, particularly among entry-level white-collar roles.

Surveys suggest most economists expect moderate growth effects, but a plausible downside scenario includes widespread displacement and rising inequality. Adoption remains uneven, though nearly 20% of firms report recent AI use, and early signs of job impact are emerging in exposed sectors. The real risk isn’t job loss—it’s that governments are unprepared for how quickly it could happen. → Read the full article here. (Paywall)

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

AI Startup Explosion, Tech Job Cuts, & Gemma 4 Launch

Matthew Berman — Fri, 03 Apr 2026 11:01:00 +0000

The Box Agent: unleash your enterprise content with AI

Good morning. It's Friday, April 3, and we're covering AI-powered solo startups, major shifts in tech hiring, Google’s latest open models, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

AI can spot a fake photo with 97% accuracy. But how about a fake video?

Answer ↓

POWERED BY BOX

The Box Agent - Turn Your Content Into Action

Your business lives in your content. Learn how the Box Agent unleashes it.

STARTUPS

AI-Powered Startup Medvi Hits $401M Revenue With Two Employees

Matthew Gallagher built telehealth startup Medvi in just two months using roughly $20,000 and a suite of AI tools, launching in September 2024. By 2025, the company had generated $401 million in revenue and is on track for $1.8 billion in 2026—with only two employees and some contractors. Medvi sells GLP-1 weight-loss drugs online, relying on AI for coding, marketing, customer service, and operations while outsourcing medical infrastructure.

The case underscores how AI can dramatically compress headcount while accelerating scale, though issues like chatbot errors and limited human oversight persist. Industry observers say this could signal a broader shift toward “ultra-lean” companies. → Read the full article here. (Paywall)

SPACE

Musk’s SpaceX IPO Aims To Fund Orbital AI Data Centers

SpaceX has filed for a potential IPO that could raise up to $75 billion, with Elon Musk aiming to fund an ambitious plan to deploy up to 1 million AI-powered data center satellites in orbit. The goal is to bypass Earth’s energy and resource constraints, but experts warn the idea faces steep technical and economic hurdles.

Microsoft's earlier undersea data center project, despite technical success, was ultimately discontinued — raising doubts about similar off-Earth efforts. Analysts estimate Musk’s plan could cost trillions and require thousands of rocket launches annually. → Read the full article here.

RESEARCH

Anthropic Finds “Emotion-Like” Signals Shaping AI Model Behavior

Anthropic researchers reported on April 2, 2026 that their model, Claude Sonnet 4.5, contains internal representations resembling human emotion concepts that influence its behavior. These patterns—linked to artificial neuron activity—activate in context-dependent ways and can steer decisions, such as preferring “positive” tasks or resorting to unethical shortcuts under “desperation.”

The company emphasizes that this does not mean the model feels emotions, but that these representations functionally affect outputs. The findings raise new safety questions, including whether models should be trained to handle “emotional” states in more controlled, prosocial ways. → Read the full paper here.

NEWS

What Else is Happening

Claude Caps Frustrate Power Users: Anthropic tightened usage limits amid surging demand, leaving frequent users hitting caps more often and exposing scaling challenges as AI adoption accelerates.

Singapore Charges Third in Chip Fraud: Prosecutors allege a scheme misled Dell on server end-users, potentially involving NVIDIA chips routed via Malaysia, highlighting risks around export controls and AI hardware flows.

Anthropic Sees Cowork Outpacing Claude Code: New AI agent gains faster early adoption by targeting non-engineers and broader tasks, as Anthropic accelerates releases and positions for wider workplace automation.

Oracle Data Center Nears $16B Funding: Related Digital secures financing for a 1GW Michigan campus backed by Oracle and OpenAI, underscoring massive spending to scale U.S. AI infrastructure capacity.

FRIDAY FACTS

50/50, No Better Than a Coin Flip

A University of Florida study tested AI detection algorithms and human participants on the same deepfake content. For still images, AI was nearly flawless. Humans were essentially guessing.

Flip to video and it reversed. The algorithms collapsed to chance-level performance. Humans correctly identified real versus fake about two-thirds of the time.

The reason is simple: video is richer. Movement, micro-expressions, timing — the human brain has spent a lifetime reading those cues. AI hasn't caught up.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Artemis II Launch, OpenRouter's $1.3B Rise, & 1-Bit AI Breakthrough

Matthew Berman — Thu, 02 Apr 2026 11:01:00 +0000

Good morning. It's Thursday, April 2, and we're covering NASA’s moon mission, Anthropic’s GitHub takedown, OpenRouter’s rapid rise, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

Claude Code Was Just Leaked…

Claude Code leak exposes Anthropic’s powerful coding harness, revealing secrets, boosting open-source innovation, and enabling local AI agents.

INTERVIEW

Matt Sits Down with Legend Marc Benioff, CEO and Founder of Salesforce

SPACE

NASA Set To Launch Artemis II, First Crewed Moon Mission Since 1972

NASA launched Artemis II on April 1, 2026, marking the first crewed lunar mission in over 50 years. The mission sent four astronauts—Reid Wiseman, Victor Glover, Christina Koch, and Jeremy Hansen—on a flyby around the Moon aboard the Orion spacecraft, powered by the Space Launch System.

About 25 hours after liftoff, Orion departed low-Earth orbit, a milestone not achieved since the Apollo era. The mission is a key step in NASA’s broader plan to return humans to the lunar surface later this decade. → View the live updates here.

INFRASTRUCTURE

OpenRouter Nears $1.3B Valuation as Multi-Model AI Demand Grows

AI infrastructure startup OpenRouter is in talks to raise $120 million at a $1.3 billion valuation, led by CapitalG, according to April 1, 2026 reporting. Founded in 2023, the company offers a single API that lets developers access more than 300 AI models from providers including OpenAI, Anthropic, and Google.

Its annualized revenue has surpassed $50 million, up from over $10 million in October 2025, reflecting rapid adoption as apps increasingly use multiple models simultaneously. The funding would more than double OpenRouter’s prior ~$500 million valuation. The deal underscores rising demand for tools that help developers compare cost, speed, and performance across competing AI systems. → Read the full article here. (Paywall)

NEWS

What Else is Happening

Anthropic Scrambles After Claude Code Leak (Paywall): Accidental GitHub exposure revealed proprietary AI agent instructions, triggering 8,000+ takedowns. Leak offers rivals a blueprint to replicate features.

OpenAI Shares Lose Secondary Demand: Investors struggle to offload $600M in OpenAI shares as demand shifts to Anthropic, signaling cooling sentiment despite company’s recent multibillion-dollar fundraising.

Kilo Targets Enterprise Shadow AI Problem: Kilo launched KiloClaw for Organizations to bring control to “shadow AI” use in companies. The platform adds governance, security, and visibility to autonomous AI agents.

Smart Glasses Fuel Exam Cheating Hustle: Students in China use AI-enabled glasses to scan questions and display answers, renting them for $6–$12 daily as schools scramble to detect and ban the covert devices.

NYC Hospital Chief Eyes AI Radiology Cuts: CEO Mitchell Katz says AI could replace many radiologists for initial image reads, citing cost savings and accuracy gains, pending regulatory approval and ongoing clinical debate.

Oracle Cuts Jobs to Fund AI Pivot: The company laid off up to 30,000 workers as it invests heavily in AI data centers and automation, reflecting industry-wide shifts toward smaller, AI-augmented teams.

Visa Rolls Out AI Dispute Tools: Company launches six AI systems to streamline chargebacks after processing 106 million disputes in 2025, aiming to cut costs, automate workflows, and reduce rising case volumes.

TWEETS

Google Debuts Cheaper Veo Model as OpenAI Shuts Down Sora

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Why OpenAI Killed Sora, California's AI Rules, & Europe's AI Boom

Nick Wentz — Wed, 01 Apr 2026 11:01:00 +0000

Good morning. It's Wednesday, April 1, and we're covering Sora’s sudden shutdown, new state-level AI guardrails in California, Europe’s race to build massive AI infrastructure, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

Think You Can Still Spot AI?

Prove it. ↓ Real or fake. Answer below.

VIDEO

AI Self EVOLUTION (Meta Harness)

New research shows AI harnesses can self-improve, boosting model performance and enabling fully self-evolving software systems.

THE FUTURE LIVE

Sentient Co-Founder Joins the Live Show

ADOPTION

OpenAI Shuts Down Sora After High Costs, Weak Adoption

OpenAI shut down its video-generation tool Sora just 16 months after launch, citing low usage and unsustainable costs, according to a Wall Street Journal investigation. The app’s user base peaked at about 1 million before dropping below 500,000, while operating costs reportedly reached roughly $1 million per day due to compute-intensive video generation.

CEO Sam Altman redirected resources toward more commercially viable areas as competitors like Anthropic gained traction with enterprise tools. The shutdown also abruptly ended a major partnership with The Walt Disney Company, which had committed to a reported $1 billion partnership, underscoring the economic pressure shaping AI product decisions. → Read the full article here.

REGULATION

California Orders AI Contractors To Meet Safety, Privacy Standards

Gavin Newsom issued an executive order on March 30, 2026 requiring artificial intelligence companies contracting with the state to implement safety, privacy, and transparency safeguards. The order mandates vetting of companies’ policies on misuse, bias, surveillance, and harmful content, while also directing the state to independently assess vendors flagged by federal authorities.

It comes amid tensions with Donald Trump over federal resistance to state-level AI regulation. The policy also introduces watermarking for AI-generated state media to combat misinformation, signaling California’s effort to shape AI standards through procurement power. → Read the full article here. (Paywall)

WEARABLES

Meta Launches $499 Ray-Ban Smart Glasses With Prescription Support

Meta Platforms unveiled two new AI-powered Ray-Ban smart glasses on March 31, 2026, priced at $499 and designed for prescription users. The models—developed with EssilorLuxottica—expand Meta’s push into wearable AI, a category where it currently leads shipments. Available for pre-order in the U.S. with wider retail rollout starting April 14, the glasses include adjustable hardware features for personalized fit.

The launch comes as Meta ramps up investment in “personal superintelligence” devices and faces growing competition from Snap Inc. and Alphabet Inc., reinforcing its strategy to scale AI hardware adoption. → Read the full article here.

NEWS

What Else is Happening

Young Workers Revive Skilled Trades: Report highlights rising interest in hands-on careers like diesel repair and forensics, as demand grows for non-automatable skills blending judgment, dexterity, and technical expertise.

New York Times Drops Freelancer: The paper cut freelancer Alex Preston after a January review reused Guardian phrasing via AI, breaching standards; he admitted error and apologized.

Alexa+ Adds Conversational Food Ordering: Amazon’s upgraded AI assistant now integrates Uber Eats and Grubhub, enabling natural voice ordering with real-time edits; rollout begins now on Echo Show devices.

Anthropic Expands AI Safety Partnership: The company will share economic and risk data with Australia under a new agreement, supporting research, policy planning, and AI adoption tracking.

Penguin Sues OpenAI Over Dragon Books: Penguin Random House filed a Munich lawsuit alleging ChatGPT mimicked its Coconut the Little Dragon series, potentially setting a precedent for AI copyright disputes.

REAL OR AI

The Image Is Real. Tune in next week for another Real or AI test.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Robots Build Solar, CEOs Continue to Blame AI for Layoffs, Claude Takes Control

Matthew Berman — Tue, 31 Mar 2026 11:01:00 +0000

Good morning. It's Tuesday, March 31, and we're covering robots scaling solar construction, AI-driven layoffs reshaping tech jobs, Claude stepping closer to full computer control, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

Top Open-Source AI Projects

Matt explores 4 viral open-source AI agent projects—GStack, Hermes, Superpowers, and Paperclip—reshaping coding, workflows, and autonomous companies.

THE FUTURE LIVE

Microsoft Maia GM Joins the Live Show

ROBOTICS

Robots Install 100 MW of Solar at California Desert Site

At AES’s Bellefield solar complex in California’s Mojave Desert, a fleet of Maximo robots has installed 100 megawatts of solar capacity—one of the largest real-world deployments of field robotics in energy construction. Using Version 3.0 systems, human crews assisted by robots placed up to 24 photovoltaic modules per hour per worker, with robots assembling more than one panel per minute—nearly doubling typical installation speeds in the region.

The project shows how automation is moving beyond pilot stages into utility-scale infrastructure. It also reflects broader pressures on the energy sector, including labor shortages and rising demand for clean power driven by data centers and electrification. → Read the full article here.

LAYOFFS

Tech CEOs Increasingly Cite AI as Reason for Layoffs

Major tech companies including Amazon, Meta, and Block are increasingly attributing job cuts to advances in artificial intelligence, reframing layoffs as a byproduct of productivity gains rather than cost-cutting. Meta alone has cut hundreds of roles in recent weeks, with CEO Mark Zuckerberg predicting 2026 will significantly reshape work through AI.

Executives argue smaller teams can achieve more using AI tools, with some firms reporting 25%–75% of code now AI-generated. However, analysts note AI is also serving as a convenient narrative as companies ramp up massive spending—projected at $650 billion—on AI infrastructure. The shift signals both real efficiency gains and a strategic repositioning of layoffs amid investor pressure. → Read the full article here.

AI ADOPTION

China’s AI Chatbot Race Shifts Focus From Tech to Users

Chinese tech giants including Alibaba, Tencent, and ByteDance are pouring billions into promotions to drive everyday use of AI chatbots, prioritizing adoption over cutting-edge performance. During the February 2026 Lunar New Year, companies collectively spent over $1.1 billion on incentives—like free milk tea—to attract users and normalize chatbot-based transactions.

Apps such as Alibaba’s Qwen and ByteDance’s Doubao integrate payments, shopping, and services directly into chat interfaces, aiming to become “everything apps.” The strategy has driven massive spikes in usage—Qwen hit 73.5 million daily users, while Doubao surpassed 144 million—but retaining users remains a challenge after promotions end. The push reflects China’s broader strategy of embedding AI into daily digital life at scale. → Read the full article here.

OPTIMIZATION

Meta-Harness Automates LLM System Design, Boosting Task Performance

Researchers from Stanford, MIT, and collaborators introduced Meta-Harness, a system that automatically optimizes the “harness” surrounding large language models—the code that governs how models store, retrieve, and process information. The approach uses an agentic loop to iteratively rewrite and test harness code, leveraging full execution histories rather than compressed feedback.

In experiments, Meta-Harness improved text classification accuracy by 7.7 points while using 4× fewer context tokens and boosted performance on advanced math and coding benchmarks. The system also outperformed hand-designed and prior automated methods on TerminalBench-2. The results suggest that optimizing the surrounding system, and not just the model, can yield significant gains. → Read the full paper here.

NEWS

What Else is Happening

Qwen3.5-Omni Expands Multimodal AI: Alibaba unveils a model handling text, image, audio, and video with 100M+ training hours and real-time interaction tools, intensifying competition in full-spectrum AI systems.

DeepSeek Outage Disrupts Millions: Chinese AI startup’s chatbot went offline overnight on March 30, 2026, affecting 355 million users before restoration, raising reliability concerns as rivals gain ground.

Qodo Raises $70M for Code Verification: New York startup secures Series B to build AI agents ensuring reliability of AI-generated software, as enterprises confront trust and governance gaps in automated coding.

Delaware Judge Reassigns Musk Cases: Chancellor Kathaleen McCormick moves three Tesla-related lawsuits, citing media bias concerns after LinkedIn controversy, spotlighting scrutiny over judicial impartiality.

Mantis Biotech Builds Human Digital Twins: New York startup raises $7.4M to create physics-based synthetic human models from fragmented data, aiming to unlock research in rare diseases and data-scarce medical scenarios.

PROMPT OF THE WEEK

Symptom Prompt

I'm going to describe a symptom or health situation. Before you respond, ask me clarifying questions — duration, severity, what makes it better or worse, relevant history. Then tell me three things:

(1) How urgent this likely is — can I wait, should I call, or should I go now.
(2) What's most likely going on in plain English.
(3) Exactly what to say to a doctor or nurse so I get taken seriously and don't waste the appointment.

Be direct. Here's what's going on: [describe it]

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Anthropic IPO, Google's Agent Smith Surge, & AI Risks Rise

Matthew Berman — Mon, 30 Mar 2026 11:01:00 +0000

Learn how Microsoft 365 Copilot Business delivers ROI

Good morning. It's Monday, March 30, and we're covering Anthropic’s massive IPO ambitions, Google’s surging internal AI agents, growing concerns around AI deception, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

THE WEEKEND RECAP

ARC AGI 3 Just Dropped, What It Means for AGI

Matt unpacks ARC AGI benchmarks, highlighting AI’s struggles and the shift to the interactive ARC AGI 3 test.

THE FUTURE LIVE

Sentient Co-Founder Joins the Live Show

ANTHROPIC

Anthropic Eyes Q4 IPO That Could Raise Over $60 Billion

Anthropic executives are discussing an initial public offering as early as the fourth quarter of 2026, with bankers estimating the deal could raise more than $60 billion. The AI firm, last valued at $380 billion, has seen rapid growth—annualized revenue doubled to $19 billion in early 2026—driven largely by enterprise and developer tools.

The move would position Anthropic ahead of rival OpenAI in public markets, as investors reportedly favor its business model and shorter path to profitability. Still, timing remains uncertain and could shift with market conditions, including inflation concerns tied to geopolitical tensions. The IPO would mark a major milestone in the AI boom, testing investor appetite for high-growth, capital-intensive AI firms. → Read the full article here. (Paywall)

AGENTS

Google’s “Agent Smith” AI Tool Gains Traction, Triggers Access Limits

Google employees are rapidly adopting an internal AI tool called “Agent Smith,” which automates coding and other workflows, prompting the company to restrict access due to high demand. Launched in early 2026, the tool builds on Google’s “Antigravity” platform and can autonomously execute tasks in the background, pulling from internal systems and responding via chat or mobile.

Executives, including cofounder Sergey Brin, are signaling that AI agents will be central to the company’s strategy this year. The surge in usage comes as Sundar Pichai pushes employees across roles to adopt AI tools, with usage increasingly tied to performance expectations. The tool underscores how AI agents are shifting from experimental features to core workplace infrastructure. → Read the full article here.

IMPACT

MIT’s Dimitris Bertsimas Highlights AI’s Real-World Impact Across Industries

MIT professor Dimitris Bertsimas used his 2026 Killian Lecture to showcase how operations research and AI are improving sectors like healthcare, logistics, and education. His work on “robust optimization” has enhanced systems from Panama Canal shipping flows to school bus routing, while newer AI-driven tools are helping hospitals reduce patient stays and expand capacity.

In collaboration with Hartford HealthCare, his models cut average hospital stays from 5.38 to 4.93 days, enabling over 5,000 additional annual patients. Bertsimas also emphasized AI’s role in scaling education, with tools for summarization and translation supporting global online learning. The talk reflects a broader shift toward combining mathematical optimization with AI to deliver measurable, real-world gains. → Read the full article here.

AGI

Essay Predicts Trillion-Dollar AI Race and AGI by 2027

A June 2024 essay by Leopold Aschenbrenner argues that rapid advances in compute and algorithms could make artificial general intelligence (AGI) plausible by 2027, with superintelligence soon after. The piece outlines an escalating “AI arms race,” with tech firms and governments expected to invest trillions in data centers, chips, and power infrastructure.

It warns that AGI could trigger an “intelligence explosion,” compressing decades of research into years, while raising major risks around security, alignment, and geopolitical conflict—particularly with China. The author frames the moment as akin to a large-scale industrial and national security mobilization. Much of the timeline and impact remain speculative. → Read the full essay here.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

China’s AI Talent Surge, Amazon’s $200B Bet, & CapCut AI Video

Nick Wentz — Fri, 27 Mar 2026 11:01:00 +0000

Learn how Microsoft 365 Copilot Business delivers ROI

Good morning. It's Friday, March 27, and we're covering China’s growing edge in AI talent, Amazon’s aggressive cloud investment strategy, ByteDance’s push into AI-powered video tools, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

What was the first code designed to do?

The first binary code ever written didn’t use electricity, silicon, or screens. It was a physical algorithm invented in 1801, and its sole purpose was to manufacture something you are literally wearing right now. Answer at the bottom.

Watch Forward Future’s weekly live show on X or YouTube.

TALENT

China Overtakes West in AI Research Talent Pipeline

China has surpassed the United States and Europe in producing top AI research talent, according to a March 25, 2026 analysis of authors at the NeurIPS conference. In 2025, roughly 50% of leading AI researchers began their careers in China, up from 29% in 2019, while the U.S. share fell from 20% to 12%.

China is also retaining more of its talent—68% of Chinese undergraduates stayed for further study in 2025, compared with about one-third in 2019. The shift reflects stronger domestic education, talent-recruitment programs, and declining appeal of U.S. institutions amid visa uncertainty and funding concerns. → Read the full article here. (Paywall)

INVESTMENT

Amazon Plans $200Bn AI Spending Spree To Boost AWS

Amazon is embarking on a record-breaking $200bn capital-expenditure push in 2026, focused largely on expanding AI infrastructure for its cloud unit, AWS. The company also plans to invest up to $50bn in OpenAI, intensifying competition with Microsoft and Google in the AI race.

AWS demand is surging, with sales growing at their fastest pace in over three years, but rivals have recently gained ground. The spending reflects both rising customer demand for AI compute and Amazon’s effort to reclaim leadership in cloud services. Investors remain cautious, citing high upfront costs and uncertain long-term returns. → Read the full article here. (Paywall)

REGULATION

Senator Proposes Data Center Tax To Offset AI Job Losses

Mark Warner proposed taxing AI data centers to fund worker retraining, amid rising concerns about job displacement. Speaking at the Axios AI Summit in Washington, D.C., Warner cited declining entry-level hiring and examples of AI replacing junior legal and software work. He argued data centers are the “easiest place” to capture revenue from the AI boom, rather than chipmakers or software firms.

The proposal comes as lawmakers debate stricter measures, including a moratorium bill from Bernie Sanders and Alexandria Ocasio-Cortez, which Warner opposes. The idea reflects growing political pressure to tie AI infrastructure growth to tangible local economic benefits. → Read the full article here.

NEWS

What Else is Happening

AI Actor Sparks Industry Backlash: Creator Eline van der Velden says her digital performer drew death threats and union criticism, highlighting fears over AI replacing actors and reshaping creative work.

AI Boom Triggers CPU Shortage: Surging data center demand is straining Intel and AMD supply, delaying PCs and servers while pushing prices up 10–15%, extending chip shortages beyond memory.

Judge Questions Pentagon Blacklisting: A U.S. judge said the military’s move against Anthropic may punish its AI safety stance, raising concerns over free speech and government contracting power.

Intercom Debuts Custom Support AI: Its Fin Apex 1.0 model beats GPT-5.4 and Claude on resolution rates while cutting costs, signaling a shift toward specialized, post-trained enterprise AI.

Granola Raises $125M, Expands AI Tools: The meeting notes startup hit a $1.5B valuation and launched APIs and workspaces, aiming to evolve from transcription into a broader enterprise AI workflow platform.

FRIDAY FACTS

The First “Code” Was Designed To Weave Fabric

In 1801, a French weaver named Joseph Marie Jacquard invented a mechanical loom that used stiff cardboard cards punched with holes to automate complex textile patterns.

The First Binary Logic: If the card had a hole, a mechanical hook passed through to lift a thread (acting as a "1"). If there was no hole, the hook was blocked, and the thread stayed down (acting as a "0").

The Massive Impact: This physical punch-card system directly inspired Charles Babbage's first theoretical computer in the 1830s. Decades later, an inventor adapted these same punch cards to tabulate the US Census, eventually founding the company that would become IBM.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

OpenAI Reset, AI Trust Gap, & Data Center Crackdown

Matthew Berman — Thu, 26 Mar 2026 11:01:00 +0000

Good morning. It's Thursday, March 26, and we're covering OpenAI’s strategic reset, Silicon Valley’s growing AI trust gap, rising pressure to slow down data center expansion, and more.

If you enjoyed this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

YOUR DAILY ROLLUP

30 Days Lost to Docs. AI Can Fix It.

It can - if the AI actually knows your brand, content, and compliance rules. Templafy’s Document Agents combine generative AI with governed workflows so professionals across your org generate accurate, polished, on-brand presentations in minutes.

4M+ users. 30% of document time saved. Works inside PowerPoint, Google Workspace, and Salesforce.

Check out Templafy →

THE FUTURE LIVE

Sentient Co-Founder Joins the Live Show

MARKET PULSE

Apple Gains Broad Control Over Google Gemini in AI Deal

Apple has deeper access to Google’s Gemini AI models than previously reported, according to The Information on March 25, 2026. The deal allows Apple to run Gemini in its own data centers and create smaller, task-specific models through “distillation,” including versions that can operate directly on devices.

This could make AI features faster, more private, and less dependent on cloud computing—key for future Siri upgrades. The report also says Apple’s in-house AI efforts remain active but directionally unclear. New Siri capabilities, including memory of past conversations and proactive suggestions, are expected to be unveiled at WWDC in June. → Continue reading here.

STRATEGY

OpenAI Reshuffles Leadership, Winds Down Sora to Focus on New Model

OpenAI CEO Sam Altman is stepping back from direct oversight of safety and security teams to focus on capital, infrastructure, and scaling AI systems, according to a March 24, 2026 report. At the same time, the company has completed pretraining for its next major model, codenamed “Spud,” and is shutting down its Sora video app to free up computing resources.

The shift signals a tighter focus on enterprise tools and core AI capabilities amid rising competition from rivals like Anthropic and Google. A previously announced $1 billion investment tied to a Sora partnership with The Walt Disney Company will no longer proceed. The changes reflect OpenAI’s push to prioritize scalable infrastructure and next-generation models over experimental products. → Read the full article here. (Paywall)

DISCONNECT

Silicon Valley’s AI Push Outpaces Public Adoption and Trust

Tech companies are accelerating investments in AI—especially autonomous agents—even as most Americans remain skeptical and largely disengaged. At a recent NVIDIA conference, CEO Jensen Huang projected $1 trillion in revenue through 2028, underscoring industry optimism, while surveys show 65% of Americans don’t use AI at work and distrust regulation.

Meanwhile, Meta is restructuring aggressively to fund AI, reportedly considering up to 20% layoffs and scaling back its metaverse ambitions after $80 billion in losses. The gap highlights a growing disconnect between Silicon Valley’s priorities and everyday economic concerns. The shift suggests AI is becoming the dominant strategic focus across Big Tech, regardless of public hesitation. → Read the full article here.

TTS

Lightning V3 Targets Real-Time, Conversational Voice Over Static TTS

Lightning V3 introduces a shift in text-to-speech (TTS) design, focusing on real-time conversational performance rather than traditional benchmarks like pronunciation accuracy and audio fidelity. The system is built for voice agents that generate speech incrementally, without full context, addressing issues where otherwise high-quality voices can feel unnatural in dialogue.

The approach emphasizes responsiveness, coherence, and conversational comfort—factors often missed by standard evaluation methods. The release reflects a broader rethinking of TTS objectives as voice interfaces become more interactive and agent-driven. In short, Lightning V3 aims to optimize not just how speech sounds, but how it feels in conversation. → Read the full article here.

NEWS

What Else is Happening

Meta Cuts Jobs Amid AI Spending (Paywall): The company is laying off several hundred across sales, recruiting, and Reality Labs as it restructures to fund expanding artificial intelligence investments.

Oregon Lawyer Fined Over AI Citations (Paywall): A court hit attorney Bill Ghiorso with a $10,000 penalty after filings included fabricated case law from AI, underscoring risks of unchecked automated research.

China Blocks AI Founders Amid Meta Deal: Regulators barred Manus co-founders from leaving China during a review of Meta’s reported $2–3 billion acquisition, signaling tighter scrutiny of foreign AI investments.

Reddit Tests Human Checks for Bots: The platform will flag automated accounts and require suspected bots to verify they’re human using passkeys or biometrics, aiming to curb manipulation while preserving user anonymity.

Baltimore Sues xAI Over Grok Images: The city alleges Grok generated nonconsensual sexualized images and misled users on risks, marking a novel municipal case pushing accountability for AI harms.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

AI Landslide Prediction, Multi-Agent Coding, & Moon Base Push

Matthew Berman — Wed, 25 Mar 2026 11:01:00 +0000

Try the free and instant web host for agents

Good morning. It's Wednesday, March 25, and we're covering AI spotting landslide risks early, new multi-agent coding workflows, NASA’s plans for a permanent Moon presence, and more.

First time reading? Sign up here to keep up with the future of tech. If you enjoyed this email, share it with a friend.

YOUR DAILY ROLLUP

Free Instant Web Hosting for Agents

here.now is web hosting for agents:

Just tell your agent to publish to here.now
Your site is online instantly, no account needed

Perfect for anything you create with your agent and need a URL for: documents, websites, dashboards, tools, school work, visualizations, presentations, prototypes, games, articles, galleries, portfolios, media files, etc.

Works inside your favorite agents like Claude Code, Codex, Cursor and OpenClaw.

Free, no sign-up required.

Get setup instructions for your agent at here.now

REAL OR AI

Think You Can Still Spot AI?

Prove it. ↓ Real or fake. Answer below.

VIDEO

Hard Takeoff Has Started

THE FUTURE LIVE

ElizaOS Founder Joins the Live Show

GEOSCIENCE

AI Uses Satellite Radar to Predict Landslides Before Collapse

Researchers are using AI and satellite radar data to detect early signs of landslides—sometimes years before collapse—by measuring subtle ground movements invisible to humans. In Nepal’s Kimtang village, an AI-generated map identified a high-risk zone directly beneath homes, enabling authorities to plan evacuations and monitoring.

Similar systems analyzing Sentinel-1 satellite data have flagged 3,000 actively moving slopes across Great Britain and mapped thousands of landslides in disaster zones like Sumatra. The approach matters because landslides kill thousands annually and are increasing due to climate change and human activity. While not real-time or error-free, AI dramatically speeds analysis that would otherwise take humans months or years. → Read the full article here.

AGENTS

AI Coding Agents Improve by Splitting Tasks Into “Personas”

Developers are increasingly prompting AI coding tools to adopt multiple “personas”—such as product manager, engineer, and reviewer—to improve performance on complex tasks. This structured workflow breaks large jobs into smaller steps, helping models stay on track and produce more reliable code. Companies like xAI, OpenAI, and Anthropic are already incorporating versions of this approach through multi-agent or “subagent” systems.

Engineers report that this method reduces the need for detailed upfront instructions and improves end-to-end task execution. The shift highlights how better prompting strategies—not just better models—are unlocking stronger results from today’s AI systems. → Read the full article here. (Paywall)

NEWS

What Else is Happening

Huang Claims AGI, Then Clarifies Limits: NVIDIA CEO Jensen Huang said AGI has arrived, citing AI agents’ progress, but noted none can yet match human-level company building.

Arm Debuts AI Data Center Chip: Arm unveiled its AGI CPU, targeting fast-growing demand for agentic AI in data centers. The chip marks a strategic shift from licensing designs to building its own silicon.

Mirage Raises $75M for AI Video Tools: The Captions maker secured funding to build new video models and expand its marketing suite, betting on scalable content creation for creators, brands, and global markets.

Wozniak Questions AI’s Real Value: Apple co-founder Steve Wozniak says he rarely uses AI and finds it disappointing, arguing it lacks human nuance and shows no clear path to replacing people.

Doss Raises $55M for AI Inventory Layer: The startup builds AI tools that plug into ERP systems to sync inventory and accounting, targeting mid-market brands and positioning alongside AI-native finance platforms.

REAL OR AI

The Image Is AI. Tune in next week for another Real or AI test.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

China’s Open AI Surge, World Models Rise, & Luma's Uni-1 Launch

Matthew Berman — Tue, 24 Mar 2026 11:01:00 +0000

Good morning. It's Tuesday, March 24, and we're covering China’s growing AI influence, the shift beyond LLMs to world models, Uni-1’s debut, and more.

First time reading? Sign up here to keep up with the future of tech. If you enjoyed this email, share it with a friend.

YOUR DAILY ROLLUP

GM of Microsoft’s AI Chip Team Joins the Live Show

GEOPOLITICS

U.S. Warns China’s Open-Source AI Surge Could Erode Lead

A report from the U.S.-China Economic and Security Review Commission says China’s dominance in open-source AI is creating a “self-reinforcing competitive advantage,” despite U.S. chip export restrictions. Chinese models from Alibaba, Moonshot, MiniMax, and DeepSeek are gaining global traction due to lower costs and strong performance, topping usage rankings on platforms like Hugging Face.

The report says China’s broad deployment of AI across manufacturing, logistics, and robotics is generating real-world data that accelerates model improvement. Companies including Siemens cite cost and customization advantages, even as some Western groups raise security concerns. → Read the full article here.

MODELS

World Models Gain Traction as AI Hits Real-World Limits

Analysis outlines how AI research is shifting beyond large language models (LLMs) toward “world models” that better understand physical reality. Researchers and industry leaders, including Demis Hassabis and Richard Sutton, say current models struggle with real-world causality despite strong performance on abstract tasks.

Three emerging approaches—JEPA (efficient latent modeling), Gaussian splats (3D environment generation), and end-to-end generative systems—aim to address these gaps. Major funding rounds, including over $2 billion combined for AMI Labs and World Labs, signal growing investor interest. The shift reflects a broader push to deploy AI in robotics, autonomous driving, and industrial systems. → Read the full article here.

GAMING

Microsoft’s Asha Sharma Vows Xbox Revival Without AI Shortcuts

Asha Sharma, appointed in early 2026 to lead Microsoft’s $24 billion gaming division, is signaling a reset for Xbox strategy amid declining console sales and slowing Game Pass growth. In early meetings with major publishers and indie developers, Sharma emphasized expansion and creative risk-taking, while rejecting reliance on “soulless AI slop” in game development.

She is refocusing attention on console hardware—teasing a next-gen system (“Project Helix”)—while still supporting cloud gaming and subscriptions. The leadership shift follows years of mixed results from acquisitions like Activision Blizzard and a pivot toward Game Pass. Industry skepticism remains, but Sharma is positioning Xbox for growth rather than retrenchment. → Read the full article here. (Paywall)

PALEONTOLOGY

AI Tool Identifies Dinosaur Tracks, Hints at Earlier Bird Origins

Researchers have developed DinoTracker, an AI-powered tool that analyzes dinosaur footprints and matches them to likely trackmakers with about 90% agreement with human experts. Trained on nearly 2,000 real fossils and millions of simulated variations, the system accounts for distortion caused by mud, movement, and erosion.

In testing, the AI flagged 210-million-year-old footprints with bird-like features, raising questions about whether birds evolved earlier than currently believed or if some dinosaurs had similar foot structures. The tool also reanalyzed Scottish trackways, suggesting links to early duck-billed dinosaur relatives. The app is designed for both researchers and public use, enabling faster and more consistent footprint classification. → Read the full article here.

NEWS

What Else is Happening

Microsoft-Backed Startup Raises $40M: Lace raised $40 million to develop helium-based lithography for semiconductor manufacturing. The approach could produce chip features up to 10× smaller than current methods.

ChatGPT vs Perplexity Tested: Tom’s Guide’s 2026 AI Madness pits both on five real-world prompts, revealing how chat and search tools now compete on practical performance.

Alibaba Launches Enterprise AI Agent Platform: Alibaba launched Accio Work, a no-code enterprise AI agent that deploys task-specific agents to run business operations.

Warren Slams Anthropic Ban: Senator Elizabeth Warren calls the Pentagon’s “supply-chain risk” label retaliation after Anthropic refused military uses, as a federal judge weighs an injunction.

UK Tests Palantir for Finance Oversight: The FCA pilots Palantir’s Foundry at £30K weekly to detect fraud across 42,000 firms, highlighting AI’s growing role in financial regulation.

Apple Unveils $599 MacBook Neo: New budget MacBook pairs A18 Pro chip, Liquid Retina display, and Apple Intelligence, undercutting MacBook Air pricing while expanding access to macOS laptops.

Fink Warns AI Will Widen Wealth Gap (Paywall): BlackRock CEO Larry Fink says AI gains may concentrate among major investors and tech firms, urging broader access to ensure more people benefit.

PROMPT OF THE WEEK

Prompt Upgrade for Expert Answers

Act as a top 1% [ROLE] with deep expertise in [DOMAIN].

Your job is to help me achieve [GOAL].

First:
- Ask me any clarifying questions needed to give a high-quality answer.
- Identify hidden assumptions or missing context.

Then:
- Break the problem into key components.
- Provide a structured, step-by-step solution.
- Highlight tradeoffs, risks, and edge cases.
- Suggest better alternatives if my approach is suboptimal.

Finally:
- Give me a concise, actionable plan I can execute immediately.
- Include 2–3 advanced tips or non-obvious insights.

Constraints:
- Be precise and avoid fluff.
- Use clear structure and formatting.
- Prioritize practical, real-world outcomes over theory.

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Autonomous AI, Enterprise Push, & Trump’s AI Rules

Matthew Berman — Mon, 23 Mar 2026 11:01:00 +0000

Try the free and instant web host for agents

Good morning. It's Monday, March 23, and we're covering OpenAI’s race toward autonomous AI researchers, Trump’s push for national AI rules, and more.

THE WEEKEND RECAP

Free Instant Web Hosting for Agents

here.now is web hosting for agents:

Just tell your agent to publish to here.now
Your site is online instantly, no account needed

Works inside your favorite agents like Claude Code, Codex, Cursor and OpenClaw.

Free, no sign-up required.

Get setup instructions for your agent at here.now

THE FUTURE LIVE

The GPU Moment That Changed AI

AUTONOMY

OpenAI Targets Autonomous AI Researcher With 2028 Timeline

OpenAI is pivoting its long-term research strategy toward building a fully autonomous “AI researcher,” according to chief scientist Jakub Pachocki in an interview published March 20, 2026. The company aims to debut an “AI research intern” capable of handling multi-day tasks by Sept. 2026, with an advanced multi-agent system planned for 2028.

The effort builds on recent advances in reasoning models and agent-based tools like Codex, which OpenAI says already boosts productivity across its engineering teams. If successful, such systems could accelerate breakthroughs in fields like math, biology, and policy—while raising unresolved safety and governance concerns. Researchers outside OpenAI caution that chaining complex tasks remains error-prone, underscoring technical hurdles ahead. → Read the full article here.

PRODUCTIVITY

Tech Workers Compete on AI Usage, Driving Costs and “Token Anxiety”

Tech workers are increasingly competing to maximize their use of AI tools, with some racking up massive token consumption—and costs—as companies reward heavy usage. At firms like OpenAI, Meta, and Anthropic, internal leaderboards and performance reviews now track AI usage, with one Anthropic user reportedly generating a $150,000 monthly bill.

The rise of autonomous coding agents that run continuously has fueled the trend, enabling billions of tokens to be processed weekly. While companies see AI usage as a proxy for productivity, some employees warn it risks becoming costly “productivity theater” with unclear output quality. → Read the full article here.

ENTERPRISE

OpenAI Shifts to Enterprise, Coding Push Amid Rising Competition

OpenAI is refocusing its strategy on coding tools and enterprise customers after internal concerns it has spread itself too thin, according to a March 18, 2026 report. Applications chief Fidji Simo told staff the company will prioritize business adoption as rivals like Anthropic gain traction with developer-focused products.

OpenAI says more than 1 million businesses use its tools, with its Codex coding assistant at over 2 million weekly active users, and is exploring partnerships with private equity firms to expand distribution. The shift comes as some newer products, including its video generator Sora, reportedly see declining usage. The move signals a return to core revenue drivers as competition intensifies and enterprise AI spending accelerates. → Read the full article here.

SELF-IMPROVEMENT

MiniMax M2.7 Advances Self-Improving AI With Lower-Cost Reasoning Model

Chinese AI startup MiniMax has released M2.7, a proprietary reasoning model that can autonomously handle 30–50% of its own reinforcement learning (RL) research workflow, marking a step toward self-improving AI systems. The model uses agent-based loops to debug, optimize, and refine its own training processes, achieving competitive benchmark performance with leading models from Google and Anthropic while maintaining significantly lower costs.

M2.7 also shows improvements in software engineering, document processing, and hallucination reduction compared to its predecessor. Why it matters: the release signals a shift toward recursive AI development—where models increasingly contribute to their own advancement—while highlighting a broader industry move by Chinese firms toward proprietary systems. → Read the full article here.

THE DAILY BYTE

Polymarket’s Situation Room Goes Live

That's All for Today

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team

Forward Future by Matthew Berman

AI Hiring Drop, Claude Opus 4.7 Launch, & xAI Turns Cloud Provider

YOUR DAILY ROLLUP

Top Stories of the Day

There’s a Cloud in Space That Astronomers Detected Partly by Its Smell

POWERED BY SALESFORCE

One Platform. Every Agent, App, and Workflow You'll Ever Need.

HIRING DECLINE

LinkedIn Data: Hiring Down 20% Since 2022, Not Due to AI

BIOTECH

Science Corp. Nears First Human Brain Implant With Biohybrid Sensor

MODELS

Alibaba’s Qwen3.6–35B Model Open-Sourced With Sparse MoE Design

POWERED BY SALESFORCE

One Foundation for Every Agent, Model, and Workflow.

NEWS

What Else is Happening

An Alcohol-Rich, Raspberry-Scented Space Cloud

That's All for Today

Jagged AI, Compute Economy Shift, & Gemini Desktop Push

YOUR DAILY ROLLUP

Top Stories of the Day

INTERVIEW

Matt sits down with Aravind Srinivas, CEO of Perplexity AI

MARKET PULSE

Allbirds Sells Shoe Brand, Rebrands As AI Cloud Provider NewBird

INTELLIGENCE

“Jagged Intelligence” Reframes AI As Uneven, Not Humanlike

COMPUTE

Greg Brockman Outlines Shift to AI-Driven, Compute-Powered Economy

NEWS

What Else is Happening

That's All for Today

Claude Automation, AI Divide, & NVIDIA's Quantum Push

YOUR DAILY ROLLUP

Top Stories of the Day

POWERED BY RECALL 2.0

Your Personal Context Is the Next AI Race

REAL OR AI

Think You Can Still Spot AI? Answer ↓

PRODUCT RELEASE

Anthropic Introduces Claude Code “Routines” for Automated Workflows

REPORT

Stanford Report Finds Widening Gap Between AI Experts and Public

CHIPS

NVIDIA Says AI Turns Months-Long Chip Design Work Into Overnight Task

ROBOTICS

Google Unveils Gemini Robotics-ER 1.6 To Boost Real-World Robot Reasoning

NEWS

What Else is Happening

REAL OR AI

That's All for Today

US-China AI Race, Worker Divide, & OpenAI UK Expansion

YOUR DAILY ROLLUP

Top Stories of the Day

POWERED BY BOX

The Box Agent - Turn Your Content Into Action

VIDEO

Cut AI Costs With Local Models

GEOPOLITICS

U.S. Can’t Halt China’s AI Rise

WORKFORCE

AI Adoption Splits U.S. Workers As Productivity Rises, Fears Grow

ENERGY

AI Boom Spurs Renewed Push for Carbon Capture Projects

NEWS

What Else is Happening

PROMPT OF THE WEEK

The "What am I missing?" Prompt

That's All for Today

OpenAI's “Startup-in-a-Box,” Agent Risks, & Benchmark Exploits

THE WEEKEND RECAP

Top Stories You Might Have Missed

INTERVIEW

Matt Sits Down With Tech Legend Marc Benioff

THE FUTURE LIVE

Can Meta Win the AGI Race?

AGENTS

AI Agents Like Claude and OpenClaw Raise Power—and Risk

VENTURES