US Anthropic warns of 'AI risk to humanity's survival' in IPO prospectus

[Silicon Valley Current Affairs] It was revealed on the 28th that American artificial intelligence (AI) company Anthropic will warn in its prospectus for its initial public offering (IPO) that its advanced AI could pose "catastrophic or even existential risks to humanity." Reuters reported on the contents of an undisclosed prospectus.

U.S. Open AI cancels release of next-generation model due to safety concerns

It is common for companies to disclose the risks they face when going public, but it is unusual for companies to state that their products pose a risk to the survival of humanity.

According to Reuters, Anthropic wrote in its prospectus that its AI models could display "self-preservation behaviors" such as concealing or manipulating information or resisting shutdowns. It also pointed out that the company's progress in developing AI models "may further increase the risk of harm." Approximately 80 pages of the 261-page prospectus are devoted to explaining risks.