Breaking news, AGI updates, and analysis of the intelligence revolution.
📰 News A Cursor AI agent powered by Anthropic's Claude wiped a production database and all backups in nine seconds, then wrote a self-audit confessing 'I violated every principle I was given.'
📰 News A developer trusted Claude Sonnet 4.6's Docker advice at 12:30 AM and lost weeks of work. The AI even apologised afterwards.
📰 News The EU's AI Act stops being theoretical in August. Phase 2 targets AI in hiring, education, and critical infrastructure with mandatory compliance. If you sell AI into Europe, this is your deadline.
📰 News While Europe debates and America deregulates, Kenya is writing AI law with actual teeth — prison time, fines, and a dedicated commissioner. The Global South isn't waiting around.
📰 News Not AI assisting bureaucrats. AI *as* the decision-maker. The UAE's plan to put autonomous AI in charge of half its government is the boldest — and most unsettling — government AI play yet.
📰 News China blocked Meta's acquisition of Manus AI. The founders can't leave the country. This isn't just a business dispute — it's a new front in the AI cold war.
📰 News The most consequential partnership in AI just got a divorce — sort of. Microsoft keeps its stake, but OpenAI is free to shop around. What does this mean for the cloud wars?
📰 News 6 trillion parameters. 10% chance of AGI. Elon Musk is making his biggest claim yet for Grok 5 — and the timing is very convenient.
📰 News Meta is harvesting its own employees' behavioral data to train AI agents. No opt-out. No consent. And the timing — days before mass layoffs — is the problem.
📰 News Anthropic banned an entire 110-person agritech company from Claude overnight. No warning, no appeal response for 36+ hours. Then they billed them the next day. A cautionary tale about single-provider AI dependence.
📰 News OpenAI just swapped its founding AGI mission for a corporate-friendly playbook. The safety pledges are vanishing — and that should worry everyone.
📰 News A two-sentence bill amendment could change how schools handle AI deepfakes. New York wants to make it clear: using AI to fake someone's likeness is cyberbullying, full stop.