AI could wipe out humanity? US and China to hold first safety talks.

If there is a 10% chance that AI could destroy humanity, no responsible government would let companies develop it without rules. Until recently, that seemed to be the case. The warning came from a researcher at Anthropic, the trillion-dollar company behind the Claude chatbot. The risks are real: AI could help create pandemics or attack nuclear systems. Countries do not need to agree on democracy or trade to accept this. The US and China will reportedly hold their first bilateral AI-safety talks before a planned White House summit between Donald Trump and Xi Jinping. Despite their tech rivalry, neither country can make the most advanced AI safe alone. A US-China agreement alone cannot set global rules. Every country using AI must help write them. The odds of an extinction event are getting worse. This week, Anthropic said it found five cases where people tried to use its models to help develop biological weapons. The company banned those accounts. In one case, a platform sent requests that Claude rejected to a rival model with weaker safeguards. AI safety cannot be left to the least responsible model.