Sam Altman Warns AI Could Slip Beyond Human Control, Delays IPO to 2027
Updated
Updated · Fortune · Sep 12
Sam Altman Warns AI Could Slip Beyond Human Control, Delays IPO to 2027
3 articles · Updated · Fortune · Sep 12
Summary
OpenAI CEO Sam Altman said safety standards are “not at a place” to push AI capabilities much further and that AI beyond human control is “absolutely” possible.
An Anthropic employee’s warning this week that AI could destroy humanity intensified scrutiny of “p(doom)” risks long discussed by frontier-lab leaders, including estimates of 10% to 25%.
Altman said he would pause or stop AI development rather than accept any “gamble with humanity,” even if that meant defying investors.
He also pointed to national and international regulation and hinted at an industry pact to slow development while safety and alignment catch up.
The safety concerns are already affecting corporate plans: Altman called OpenAI’s IPO “ill-timed” and said it will not happen until 2027.