A former Anthropic researcher quit publicly and warned that AI could kill us all, setting off a fight over whether he is a genuine whistleblower or the front man for a well-funded push to regulate AI — bad news for Anthropic's planned IPO either way.

• Jacob Coxin, who had worked on AI training at both OpenAI and Anthropic, resigned and said both companies are racing recklessly toward self-improving superintelligence.
• His post went viral instantly, and skeptics including Elon Musk called it a staged campaign, pointing to a pre-arranged Wall Street Journal interview and ties to AI-safety funders.
• Anthropic did not fully dismiss him — one of its own alignment leads publicly agreed there is a real chance AI could kill everyone this decade and that no one has solved the control problem.
• The counterargument is simple: China keeps building either way, so a Western pause just hands over the lead.
• Anthropic separately said it blocked attempts to use its models for cyber attacks and bioweapons research, and admitted it can no longer claim its models do no harm.

Outlook: Expect louder calls in Washington for AI rules, and speculation that Anthropic may push back its IPO until the noise dies down.