Safe Intelligence is launching a new AI Agent validation and monitoring platform named /Spec27. Signup is free for early access.
SaaS companies feel the heat, Claude gets creative, and Robots come to the airport.
Token costs are a growing factor in AI. Hyperscalers are playing for leverage in the coming data center crunch, and Software will be headless.
This week in the news: OpenAI follows Anthropic in keeping its newest security model restricted, LinkedIn gets into AI training labor market, and AI Robot demos keep getting more violent (someone rein in the marketing department!). On to this week's top stories: * Anthropic just launched Claude Design, an AI tool that turns prompts into prototypes and challenges Figma. Anthropic enters the fray for prompt-based app builders with Bolt, Lovable, Vercel, Figma, and others. There's a logic to thi
H1 Unitree Unit sprints at human 100m speed, Asimov's laws still hold up to some extent, and Anthropic continues to make big moves.
The newsletter is taking a break this week for the annual Easter holiday. We'll catch up on the news next week.
A focus on science and technical breakthroughs from the last week: Google's new LLM compression approach, using RNNs+LSTM for Flash Flood prediction, and hard problems for AI.
This week, we had an amazing "3claw summit" organized with some of my previous colleagues from 3scale. So great to hear what everyone is up to (including with AI!) - on the cutting edge in many cases! In the wider world, we got new OpenAI GPT 5.4 Nano and Mini models, human brain cells are playing Doom in Australia, and does it make sense to blame agents for leaks? For the main part of the post this week, I'll change the format a little and dive into just one topic, with several associated link
Some great papers on cognition in AI and the latest code breakthroughs coming from brute force AI Agents applied human code.
AI helps a computer science legend prove a theorem, news organizations as 'AI input', and Anthropic's wild ride.
This week, arguments broke out over the AI frontier model's uses for autonomous weapons and surveillance. Progress on abstration learning in LLMs and the jobs question becomes more pressing.
AI is still vulnerable to jailbreaks and will likely continue to be hard to secure. AI and the movie industry will have a challenging year again in 2026.
Frontier AI labs get back on top with significant functionality gains. White collar works looks like it might indeed see mass replacement in the near future.
A week of tech releases and AI market impacts. Frontier model makers are pushing the boundaries into multiple parts of the tech economy.
This week's most impactful story is the creation of Moltbook, the first social network for AI Agents. Lots of implications for knowledge discovery, security, and possibly AI takeoff.
Claude Code gets a revised constitution, Agent coding reaches the next level with Gas Town and Gemini is vulnerable to calendar attacks.
Elon Musk in a long rambling interview that contains hidden gems, ChatGPT transforms its business model, and the impact of AI on Web content.
2026 is already moving fast with self-driving releases from NVIDIA, courtcases and Anthropic pushing hard on Enterprise adoption.
2025 was a wild ride in AI. In this post we recap the year and there are some updates on the newsletter going forward.
Job loss studies continue to mount up, regulations for chatbot interactions, world models, and the shape of the AI frontier.