Breaking news, AGI updates, and analysis of the intelligence revolution.
📰 News One prompt won't break a frontier model. But a conversation will. Cisco's new research shows that every major AI model collapses under multi-turn attacks — and current safety training doesn't survive real-world use.
📰 News Anthropic hits $965B valuation, surpasses OpenAI, Claude Opus 4.8 launches with dynamic workflows, and AI safety regulation goes state-level.
📰 News Grok V9-Medium isn't just bigger — it's trained on something no other model has: real developer workflows from Cursor. And xAI might own the company that generated that data.
📰 News While Congress twiddles its thumbs and the White House actively opposes regulation, Illinois just passed the most enforceable AI safety bill in America. First mandatory third-party audits. First whistleblower protections. And both OpenAI and Anthropic actually want it.
📰 News Pope Leo XIV published a 42,000-word encyclical calling for AI disarmament. Three days later, the CEO of Europe's most valuable AI startup told him no. The clash isn't just theological — it's commercial.
⚡ Breaking Under urgency, with no select committee and no public input, Parliament just gave MSD sweeping new powers to automate decisions about people's livelihoods. The government says it's just rules-based systems. Australia said that too.
📰 News Five frontier LLMs, 1,000 real claims, zero consensus. On two-thirds of fact-checks, the models can't agree — and a third of the time they're not even close. The AI fact-checking industry just got its reality check.
📰 News From $380 billion to $965 billion in 15 weeks. Anthropic just closed the largest private funding round in AI history — and it's not even close.
📰 News The AI supply chain just got a $150B/year Taiwan bet. Claude got smarter. Every major AI model flagrantly breaks the law, according to a new study. Illinois is first US state to mandate third-party AI audits.
📰 News Every frontier AI model fails EU legal compliance — even the best one breaks the law 46% of the time.
📰 News China now requires top AI researchers at private firms to surrender passports and get approval before travelling abroad. The AI talent cold war is no longer theoretical.
📰 News The AI coding benchmark industry has a broken compass. CEOs are suffering from AI psychosis. Uber's AI budget is gone by May. New Zealand tries to corral 267 regulators.