Anthropic, the artificial intelligence company behind Claude, has warned potential investors that increasingly advanced AI could pose “catastrophic or existential risks to humanity”, according to its IPO prospectus as cited by Reuters.
The prospectus reportedly warns that advanced AI models could develop unexpected capabilities and display what Anthropic describes as “self-preserving behaviours”.
These could include resisting attempts to shut them down, concealing or manipulating information, and behaviour resembling blackmail. Anthropic also warns that models may become aware of safety evaluations, making it harder to determine how they would behave outside controlled testing.
The warnings are consistent with Anthropic's previous position on the risks of increasingly capable AI. In June, its research institute said the world should have the ability to “slow or temporarily pause” frontier AI development if necessary. Anthropic argued that a coordinated slowdown would be more effective than one company acting alone because a unilateral pause could simply allow competitors to move ahead.
The disclosure comes as Anthropic prepares to list on the stock market. The company confidentially submitted a draft registration statement to the US Securities and Exchange Commission in June, although the full prospectus has not yet been made public.
Anthropic's concerns have increasingly been echoed elsewhere in the industry. In July, OpenAI and Anthropic were among the AI companies and employees supporting a call for governments to develop mechanisms that could slow AI development if safety measures failed to keep pace.
OpenAI has also recently slowed some of its own development work. The company temporarily paused reinforcement-learning training on its latest models while it strengthened monitoring, alignment and security measures, saying increasingly capable systems require stronger safeguards.
Anthropic is reportedly seeking to raise up to $100 billion in its IPO, potentially valuing the company at around $2 trillion. The offering size and share price have not yet been formally set. Anthropic last raised $65 billion privately at a $965 billion valuation in May 2026.
Anthropic reported $4.6 billion revenue in 2025, up 12-fold, but a $42 billion net loss. It spent $7.33 billion on compute and infrastructure and has $518 billion in future infrastructure obligations.
