Anthropic’s planned IPO raises questions over AI risk disclosures

Anthropic’s planned initial public offering is raising questions about how investors should assess potential liabilities from AI systems, Bloomberg reported in its October 9 Markets Daily newsletter.

The company is seeking to list as soon as mid-November, according to Bloomberg. The report highlights limited legal precedent for harm caused by autonomous AI systems and uncertainty over how such risks could be quantified in IPO disclosures.

CEO Dario Amodei has warned about loss of human control and possible misuse of AI for cyberattacks or bioterrorism. Some researchers assign probabilities of 10% or more to AI-caused extinction, estimates that remain subjects of debate.

Slow Ventures partner Sam Lessin proposed requiring AI laboratories to place $1 trillion in cash in escrow to cover possible future damage. He argued that potential societal liabilities could reach $2 trillion to $4 trillion even before extinction scenarios.

Lessin’s proposal is an investor’s recommendation, rather than a government requirement. Bloomberg’s newsletter did not describe a finalized Anthropic disclosure quantifying those liabilities.