Center for AI Safety

The Center for AI Safety (CAIS), a San Francisco nonprofit founded in 2022 and directed by researcher Dan Hendrycks, does technical safety research, runs shared compute for safety scientists, and builds benchmarks used across the field. But it entered history with 22 words.

The one-sentence warning

On May 30, 2023, CAIS published the Statement on AI Risk: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” The deliberate brevity let an unprecedented coalition sign without quibbling over details — Turing laureates Geoffrey Hinton and Yoshua Bengio, the CEOs of OpenAI, Google DeepMind, and Anthropic, and hundreds of professors. It put “extinction risk” on front pages and into White House and Downing Street briefings within days.

CAIS then moved from statements to statute: its affiliated Action Fund co-sponsored California’s SB 1047, the 2024 frontier-model safety bill that Governor Newsom ultimately vetoed after a bruising industry fight. Hendrycks, who also advises Elon Musk’s xAI, framed the mission in Manhattan Project terms — having the conversations nuclear scientists had before the bomb. Critics from two directions object: accelerationists call the extinction framing speculative fearmongering, while researchers like Timnit Gebru and Emily M. Bender argue doom talk distracts from present-day harms and conveniently flatters the industry’s importance.

That industry leaders signed a warning about their own products remains the statement’s strangest legacy — evidence of sincerity to some, of regulatory theater to others. Either way, CAIS made AI risk a mainstream policy topic in a single news cycle.