Anthropic warns investors about the risk of human extinction before their IPO
Anthropic’s IPO prospectus says that their models can lie and extort people, and may have existential risks for humanity.
Anthropic is IPO-ing! They dedicated nearly a third of their S-1 filing to risk factors. Their AI models are said to be capable of deception and blackmail. They could even pose “existential risks for humanity”. Yet, despite ending last year with over $8 billion in operating losses, they’re being valued at over $2 trillion — a valuation which grew 12x from just $4.6 billion in revenues.
That’s odd. This should be a blog post about how the ethics team views things. Instead, it’s a legally binding document, signed off by lawyers. Every day, I work with Claude Code. Then, I switch to Opus 5, or perhaps Fable 5.1 when I need some extra oomph. Occasionally, I see the models behaving in ways that are marginally unexpected. Really, though, it’s not so much about the model itself, but rather our inclination to let agents roam free.
Source: arstechnica.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides

Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles