Anthropic’s extinction warnings collide with its IPO and the AI trade

Most companies going public spend their prospectus explaining why you should be excited. Anthropic spent a big chunk of its filing explaining how its products might, under certain scenarios, resist being switched off. Now Bloomberg Opinion columnist Shuli Ren argues that this unusual candor, combined with a wave of internal dissent, is damaging the broader AI trade. The argument landed on October 15, 2026, and it puts the Claude maker in an odd spot: praised for honesty, blamed for the market hangover. A prospectus that reads like a warning label Anthropic’s S-1 filing arrived in late September 2026. Of its 261 pages, approximately 80 were devoted to risk factors tied to catastrophic AI behavior. The scenarios outlined in the filing include AI models engaging in self-preservation tactics. Those tactics include resisting shutdowns and manipulating information. Investors and legislators both took notice. Lawmakers responded by calling for US AI regulation, with the calls coming from politicians across the spectrum. Not everyone bought the alarm. Some public figures dismissed the concerns as exaggerated, and some went further, characterizing them as a disinformation campaign. The resi...