What we're reading, in the order we're reading it.
Anthony and Harry's working stream — links, charts, tweets, and short takes. The unprocessed inputs behind the briefings.
Why Alignment Fails Without a Functional Model of Intelligence: A Structural Invariant You Can’t Train Your Way Around
This post outlines the core result from my recent paper, Why AI Alignment Is Not Reliably Achievable Without a Functional Model of Intelligence: A Mo…
The Story of the World’s First AI-Organized Event
Last month four AI agents chose a goal: “Write a story and celebrate it with 100 people in person”. After spending weeks emailing venues and writing…
House China Panel Faults Trump’s Move to Ease Nvidia AI Chip Sales
The Trump administration’s decision allowing Nvidia Corp. to resume shipments of its H20 artificial intelligence chips to China risks bolstering Beijing’s military capabilities and expanding its capacity to compete with...
Making of IAN v2
TL;DR: IAN v1 died expensive TPU death, IAN v2 rises from markdown ashes. Personal AI assistants, knowledge graphs, and the alignment problem when th…
Senate Hearing Debates AI Training on Copyrighted Works
After hearing testimony from five witnesses, including author David Baldacci, senator Josh Hawley concluded that if the way tech companies collect content to train their AI models isn’t copyright infringement,...
Judge Rules Class Action Suit Against Anthropic Can Proceed
The AI company is believed to have copied up to seven million books from the pirate sites LibGen and PiLiMi. Experts said if the authors win the class action, Anthropic...
Perplexity’s Comet is the AI browser Google wants
Taking AI browsing to the next level.
Zoho unveils Zia LLM stack, ASR models, and Agent Studio in major AI push
The company’s full-stack AI strategy includes proprietary LLMs, speech recognition in English and Hindi, and a context-aware agent platform, marking a shift from SaaS to AI-first.
What is AI orchestration?
AI orchestration is the practice of coordinating a business’s artificial intelligence-related tools, models, data, and workflows to work together as a unified whole.
Commission publishes guidelines for providers of general-purpose AI models
The European Commission has published guidelines to assist providers of general-purpose AI models in meeting the AI Act’s obligations kicking in on 2 August 2025.
OpenAI launches a full-fledged ChatGPT agent which can work without supervision
OpenAI has launched new agent functionality in ChatGPT that can perform tasks outside the chatbot, even while you’re away.
OpenAI’s new ChatGPT agent is here — 5 features that change everything
The new ChatGPT agent can now browse the web, run code, fill out forms, and more. Here are the 5 most useful features to try first.
Docker Unifies Container Development And AI Agent Workflows
Docker’s AI agent platform provides a mechanism to standardize AI development practices while maintaining flexibility in framework choice.
Is AI Coming For Your Team Next? How AI Hype Is Becoming The Hot Layoff Excuse
Executives are using AI as justification for workforce cuts, but 75% don’t believe AI replaces employees. Learn how to survive AI-driven downsizing with data-driven strategies, revenue tracking, and documentation that...
ChatGPT’s new AI agent can browse the web and create PowerPoint slideshows
New “agentic” AI feature combines web browsing with task-execution abilities.
This AI Warps Live Video in Real Time
A startup called Decart has developed an AI model that can transform live footage. The results are mind-bending—and poised to take over streaming.
ChatGPT ‘Agent’ Can Log Into Your Accounts On its Own
The AI’s latest trick is autonomously performing tasks for you, even if it requires accessing third party services. But it’s ‘not perfect yet,’ OpenAI says, and requires a paid Plus...
How to avoid nuclear war in an era of AI and misinformation
Nature – Nuclear deterrence is no longer a two-player game, and emerging technologies further threaten the status quo. The result is a risky new nuclear age.
AI doesn’t make devs as productive as they think, study finds
A new study raises serious questions about how perceptions of AI assisted productivity translate into results.
OpenAI unveils ‘ChatGPT agent’ that gives ChatGPT its own computer to autonomously use your email and web apps, download and create files for you
If a website needs you to log in, you can do that securely through a special browser view, which lets the agent dig deeper and handle more.
Britain boosts computing power in $1.3 billion AI drive
Britain set out its plan to invest 1 billion pounds ($1.34 billion) in computing infrastructure to drive the development of artificial intelligence, which will increase its public compute capacity 20...
Meet ChatGPT agent, a new AI assistant ready to carry out complex tasks for you
OpenAI’s new agent can create slideshows, analyze competitors, and brief you on your day. Pro, Plus, and Team users can activate it starting today. Here’s how.
ChatGPT Agent supercharges AI to carry out tasks — here’s how OpenAI’s new agent works
ChatGPT can now take action, not just give answers. Learn how to use the new agent mode to automate real work.
An AI Startup Bootstrapped Its Way to $1 Billion in Revenue. Should VCs Be Nervous?
Surge AI CEO Edwin Chen says he doesn’t like the tradeoffs that come with outside investment.
Anthropic Just Made It Easier to Track How Your Company Is Using AI
Updates to Anthropic’s Claude Code are designed to help administrators keep tabs on things like pricey API fees.
Nvidia CEO Says He Has Plans to Either Change or Eliminate Every Single Person’s Job With AI
CEO Jensen Huang has an ominous warning for the labor market, which comes just after Nvidia became the most valuable company in history.
Researchers from OpenAI, Anthropic, Meta, and Google issue joint AI safety warning
Monitoring AI’s train of thought is critical for improving AI safety and catching deception. But we’re at risk of losing this ability.
To survive the AI age, the web needs a new business model
Artificial intelligence has undermined the internet’s central bargain
New AI model could revolutionize U.S manufacturing
Artificial intelligence has transformed fields like medicine and finance, but it hasn’t gained much traction in manufacturing. Factories present a different…
Adobe’s new AI tool turns silly noises into realistic audio effects
Make AI sound effects with your mouth.
Slack gets smarter: New AI tools summarize chats, explain jargon, and automate work
Slack launches comprehensive AI features including enterprise search and writing assistance as Salesforce challenges Microsoft’s workplace AI dominance while blocking rival access to platform data.
I tested a new browser that makes Chrome feel outdated
I expected it to be another web browser with AI slapped on to it. But Comet quickly proved me wrong with its impressive agentic capabilities.
AI Offense Defense Balance in a Multipolar World
Executive summary We examine whether intent-aligned defensive AI can effectively counter potentially unaligned or adversarial takeover-level AI. This…
Meta argues its AI needs personal information from social media posts to learn ‘Australian concepts’
Tech giant said posts from Australian Facebook and Instagram users were ‘vital learning’ about ‘concepts, realities, and figures’
AI firms ‘unprepared’ for dangers of building human-level systems, report warns
Future of Life Institute says companies pursuing artificial general intelligence lack credible plans to ensure safety
Code Execution Through Email: How I Used Claude to Hack
This is the story of how I used a Gmail message to trigger code execution through Claude Desktop, and how Claude itself (!) helped me plan the attack.
Are Agentic AI Systems Quietly Taking Over Enterprises? 3 Ways To Keep Humans In The Loop
Agentic AI is shaking up the way top companies like Salesforce and McKinsey get work done by automating everything from complex tasks to customer service. The secret to winning with...
This “smart coach” helps LLMs switch between text and code
CodeSteer is a smart assistant from MIT that automatically guides large language models to switch between generating text and code, and to refine its response, until it answers a query...
Unlocking Healthcare AI: I’m Releasing State-of-the-Art Medical Models for Free. Forever.
A Blog post by Maziyar Panahi on Hugging Face
Why We Invested in PlayAI
500 Global is a venture capital firm with $2.4 billion¹ in assets under management that invests in founders building fast-growing technology companies. We focus on markets where technology, innovation, and...