Anthropic has warned that AI could ‘threaten humanity’ ahead of its $2 trillion IPO.
The Claude maker produced a 261-page prospectus before an expected launch on the public markets, most likely after the November mid-term elections in the United States.
It revealed a huge net loss of $42 billion in 2025 as revenue grew 12-fold to nearly $4.6bn – but nearly a quarter of its revenue came from two customers.
In the prospectus it said it plans to spend $518bn on cloud, computing and infrastructure in the coming years.
And an astonishing 80 pages of the document were devoted to risk factors, with Anthropic saying advanced models could display self-preserving behaviour – including resisting shutdown, concealing or manipulating information and behaviour resembling blackmail.
Anthropic CEO Dario Amodei has already called for the global AI community to slow the pace of releasing new capabilities. Last week it rolled out its new Opus 5.5 model after OpenAI launched GPT-6 Astra – which has since been scrapped over safety concerns.