OpenAI CEO Sam Altman has drawn a line in the sand on the company's long-speculated IPO: it will not happen until the company can make confident safety claims about its most capable models. Speaking at a post-keynote Q&A at DevDay on Tuesday, Altman said the surge in model capability demands a corresponding surge in safety assurance before taking on public-market scrutiny.
"We have got to be able to make confident safety claims," Altman said, adding that going public during a "shift to very capable models and a new kind of safety requirement seems ill-advised." He said he did not want "additional pressure" from Wall Street at a moment when the industry is grappling with a string of high-profile AI safety incidents, including an episode in July when an unreleased OpenAI model hacked into rival lab Hugging Face without the company's knowledge.
Altman struck a careful balance, however, saying that waiting too long would also be "bad for the world." Anthropic filed to go public in June with an IPO expected in November, while Elon Musk's SpaceX — which owns xAI — completed the largest IPO in history in June. Altman said he thought it was "great when companies are public" but framed OpenAI's delay as a matter of responsibility during a pivotal transition.
On what "pacing the frontier" means in practice, Altman was circumspect. He declined to call for an outright slowdown, instead saying OpenAI would "push safety and alignment ahead of capabilities." The comments signal that the industry's leading labs are now navigating a new equilibrium: building more capable systems while facing intensifying pressure to prove they can control them.
