Exclusive: Sam Altman addresses AI doomsday fears in new interview

AI doomsday talk got loud this week after an Anthropic employee warned on X that humanity could be wiped out by the technology and that frontier labs were gambling with people’s lives. The phrase “p(doom)” has been floating around the AI world for years, but the argument is no longer tucked away in research circles. It’s now public, blunt, and increasingly hard to ignore.

Against that backdrop, Fortune sat down with Sam Altman for an hour-long interview in San Francisco on Friday. The OpenAI chief talked through the risk AI could pose to humanity, how fast the models are moving, whether stronger systems can actually be controlled, and what happens if he concludes safe development is no longer possible.

Altman’s answer was not a reassurance tour. He said no gamble with humanity is acceptable, and that he would be prepared to stand up to investors if he had to pause or stop AI development. He also said safety standards are not yet at a point where pushing capabilities much further makes sense.

The other notable detail was the IPO. Altman called it “ill-timed” because of the safety concerns and said it won’t happen until 2027. He also pointed to the need for both national and international regulation, and referred to a pact he’s making with industry peers to slow down while safety and alignment work catches up with model capability.