Anthropic readies IPO marketing as researcher puts extinction odds above ten percent

Anthropic is moving toward an IPO marketing push even as one of its own researchers publicly forecasts a greater than ten percent chance that artificial intelligence wipes out humanity within ten years. The Information reports the prediction came from an active employee — a collision of apocalyptic warning and capital-markets maneuver that is hard to ignore.

The tension has a human face. Jacob Coxon quit the company roughly two months ago over safety concerns, walking away two months before his first vesting cliff. Anthropic uses a six-month cliff; Coxon had been there four. He forfeited his entire equity stake. Axios first reported the departure, a signal, if one were needed, that internal alarm runs deep enough to cost real money.

The IPO isn't optional, people close to the company say. Training and running frontier models burns hundreds of millions of dollars a year, and subscription plus API revenue still doesn't cover the full cost. If the offering happens, it will be a survival move more than a growth play.

Critics argue Anthropic's flagship "Constitutional AI" framework functions mainly as a marketing salve for regulators and investors, while the same commercial incentives that drive its rivals steer the ship. The cap table — Google, Amazon, and a roster of blue-chip venture funds — only reinforces the case that financial imperatives trump any stated safety pledge.