Anthropic reportedly warns of existential AI risks in IPO prospectus

Anthropic is telling investors that advanced artificial intelligence could pose "catastrophic or existential risks to humanity," according to reports cited by The Guardian, as the company prepares for a potential $2tn (£1.5tn) stock market flotation.

The warning is contained in the startup's IPO prospectus, which has not yet been made public, and was reported by Reuters and the Financial Times, according to The Guardian's account.

The Guardian's report also referenced admissions to investors regarding AI's "self-preserving behaviours," though the specific details of what those behaviours entail were not fully outlined in the excerpt available.

Anthropic, the developer of the Claude AI models, has previously called for a slowdown in what it has described as the breakneck pace of AI development, a position that has been echoed at various points by other companies in the sector.

The disclosure of existential risk warnings inside a company's own IPO documents underscores a broader tension in the AI industry, where firms have simultaneously promoted the transformative potential of their technology to investors while publicly acknowledging significant safety concerns about its development.

The Guardian's report did not specify a timeline for the potential IPO or confirm the $2tn valuation figure beyond describing it as a potential flotation size being reported elsewhere.