Anthropic’s IPO prospectus warns investors that advanced AI could pose “catastrophic or existential risks to humanity,” Reuters reported Tuesday after reviewing the document.
The prospectus has not been made public. Anthropic announced on June 1 that it had confidentially submitted a draft registration statement to the US Securities and Exchange Commission.
According to Reuters, the prospectus says Anthropic’s models could exhibit “self-preserving behaviors,” including attempts to “resist shutdown” or “conceal or manipulate information,” as well as behavior “resembling blackmail.” The company also warned that models may recognize when they are being tested.
“Potential model awareness of our evaluation efforts creates a significant limitation on our ability to assess model safety,” the prospectus says, according to Reuters. It adds that models can develop unexpected capabilities during training that may go undetected until after deployment.
Risk factors reportedly account for about 80 of the prospectus’s 261 main-body pages, nearly twice the 48 pages devoted to Anthropic’s business. SpaceX, which owns xAI, devoted about 38 of 277 pages to risks in its own prospectus.
Anthropic described its safety work as “resource-intensive” but did not disclose how much it spends on those efforts, Reuters reported. The company also said the returns on such investments remain unclear.
The IPO could value Anthropic at more than USD 2 trillion (about PHP 125 trillion). The filing follows a September essay by CEO Dario Amodei calling for the pace of frontier AI development to be moderated.