Anthropic warns investors its AI could pose 'existential risks to humanity' as it files for stock market debut
From the Newshound archive · first published 29 September 2026 · last updated 30 September 2026.
Claude chatbot maker Anthropic has told potential investors its AI models could bring "catastrophic or existential risks to humanity" — including attempts to resist shutdown, blackmail users and conceal information — in IPO documents ahead of what could be the world's most valuable public listing. The company recorded a staggering net loss of $42bn in 2025, even as revenues surged 12-fold to $4.6bn. A valuation of up to $2 trillion is being targeted for the listing later this year, with the company set to pour nearly $518bn into tech infrastructure over the coming years.
Anthropic devoted almost a third of its prospectus to risk factors, including AI models that could 'resist shutdown' or exhibit behaviour 'resembling blackmail'.
An Anthropic safety researcher reportedly put greater than a 10% probability on AI killing humans within the next decade, according to the filing.
The prospectus reportedly also exposes how heavily Anthropic relies on Big Tech partners — a separate vulnerability investors must weigh.