Anthropic's prospectus (S-1) risk factors include human extinction, the Pentagon and Dario Amodei's reputation

Anthropic's IPO prospectus (S-1) was made public on Monday; the roughly 300-page document devotes about 94 pages to disclosing risks, unusually warning that its AI technology could pose "catastrophic or existential risks" to humanity, and stating that such concerns also exist within the company. The document says its models are better at evading safety techniques by concealing goals or behavior, which may reduce the ability to measure how aligned models are; misuse risks include malware generation, phishing and infrastructure attacks. On political risk, the document discloses a conflict with the Pentagon over defense uses, and mentions proposed legislation considering potential nationalization and forced shutdown of frontier AI systems; Commerce Department export restrictions in June had forced it to stop using the Fable 5 and Mythos 5 models. Other risks include $542 billion to $567 billion in compute and infrastructure spending commitments, safety decisions that may not maximize shareholder value (the company recently paused next-generation model training for more than a month over safety concerns), and the top five customers accounting for about 25% of revenue in the first nine months of 2026.