📰 News Google Is Using NVIDIA's Own Playbook to Dismantle Its AI Chip Monopoly
Google's TPU stack is going external. The CUDA moat is finally under siege from the one company with enough compute to make it stick.
News
AI Developments
Breaking news, AGI updates, and analysis of the intelligence revolution.
📰 News 
📰 News Switzerland's Apertus Is the Open Model Sovereign AI Was Promising
Switzerland's Apertus model — 8B and 70B parameters, Apache 2.0, fully open from weights to training data — hit 167 points on Hacker News as the credible European alternative to closed US/China AI.
📰 News Claude Goes Down, and the Case for Open Models Gets Harder to Ignore
Claude's June 22 outage across all Opus and Sonnet variants hit claude.ai, the API, Claude Code, and Cowork. As Anthropic investigates, a blogger's 'just switch to open models' argument is gaining traction.
📰 News Granta Won't Publish Prize Winners Anymore. AI Broke the Literary Award Trust Model.
Granta will no longer publish prize winners from external competitions after an AI controversy. The magazine says it can't verify what's human anymore.
📰 News AI Agents Can Now Deploy Code Without You Signing Up. Cloudflare Just Made It Default.
Cloudflare's --temporary flag means an AI agent can deploy a Worker, iterate on it, and hand you a claim link — no signup, no token, no human in the loop.
📰 News Brands Are Quietly Replacing Influencers With AI. You're Not Supposed to Notice.
The influencer economy is being quietly replaced by synthetic personas. Creators are under NDA. There are no rules forcing brands to tell you.
📰 News The UK Is Spending £75M to Put AI Inside Every Police Force. The Results Are Already Startling.
£75M for a national AI policing centre. A kidnapping trial finished 800 hours of footage review in 3 hours. The UK is moving fast — and NZ should be taking notes.
📰 News Signal's Whittaker: AI Chatbots Are Not Your Friends, and Agentic AI Is a Backdoor
Meredith Whittaker dismantles the 'AI as companion' framing and warns that agentic AI needs your credit card, browser, and Signal — calling it a backdoor.
📰 News Trump Says Anthropic Isn't a Threat Anymore. The Export Controls Still Stand.
Political warmth from Trump at G7, but the bureaucratic machinery hasn't moved. Anthropic's $965B IPO hangs on regulatory clarity that one interview can't deliver.
📰 News A Brussels Thought Experiment Went Viral. Then Trump Banned Anthropic's Models the Next Day.
A thinktank fiction about Europe falling behind the US and China on AI has been read by EU parliament members and discussed in UK-German diplomatic channels. Its predictions are already half-true.
📰 News From PGP to Mythos: The US Government Has Never Successfully Export-Controlled Software
Three decades of failed export controls on encryption, spyware, and now AI models suggest the Anthropic ban will restrict the wrong people.
📰 News Nobel Laureate John Jumper Leaves DeepMind for Anthropic — Google's Brain Drain Is Now a Pattern
AlphaFold co-inventor and 2024 Nobel laureate John Jumper announced he's leaving Google DeepMind after 9 years to join Anthropic. His tweet hit 548K views in hours.