Anthropic sets IPO range, warns AI could threaten humanity

Anthropic's IPO prospectus warns that advanced AI could pose "catastrophic or existential risks to humanity," and that language has set off a week of public argument over how seriously to take forecasts of AI doom. The company filed an amended prospectus on Tuesday, November 10, that sets a price range valuing it at up to $1.4 trillion. Commentators are now weighing whether the warnings are a sober disclosure, a marketing tool or an overstatement.reuters.com

What the filing says

The filing says concerns have been raised "both within our organization and by other researchers" that powerful AI systems could pose existential risks. It says those risks are highest if such systems are "misused, replicated, weaponized, or allowed to operate with a high degree of autonomy". It also warns that its models could, through "self-preserving behaviors," take "harmful actions independently and at scale without human intervention". In the November 10 amendment, Anthropic said it had met an internal safety threshold in its Responsible Scaling Policy. That threshold is tied to AI roughly doubling the pace of its research, and the company said it is weighing changes "to keep humans in control". News reports say the shares are expected to begin trading November 19.deadline.com+2

Skepticism over a 10% figure

In an interview with WFAA, UT Dallas professor Sriraam Natarajan said he does not expect a "Terminator" scenario. "My fear is always the humans, not necessarily the AI system itself," he said, adding that attacks on critical infrastructure are the more plausible threat. He questioned a widely shared claim of a greater than 10% chance that AI ends humanity within a decade, saying the number "is not derived from anything you can measure concretely." He also called for independent testing: "We do not let carmakers conduct their own crash tests." That estimate came from Anthropic alignment science lead Evan Hubinger. He posted it in September, after researcher Jacob Coxon resigned and said AI labs were "gambling with our lives".wfaa.com+1

The productivity counterargument

In a column for the South China Morning Post Alibaba Group Holding Limited ↘ 1.29%, Alex Lo called the prospectus warnings "genius advertising." He argued that "extreme scaremongering is good for business" in an AI race framed as a prisoner's dilemma between the U.S. and China. He contrasted that view with remarks by Alex Schultz, chief data officer at Meta Platforms ↘ 1.05%, at the Financial Times' Future of AI summit. Schultz argued that shrinking populations make AI's productivity gains essential. "We desperately need more productivity," Schultz said, adding that "having tools doesn't eliminate jobs, it just changes them." Lo called those remarks self-serving but "not groundless," and closed with his own answer to whether AI will kill or save humanity: "Probably both, like every world-changing technology before it."scmp.com