AINEWS
In the same Fortune interview, Altman said building AI beyond human control is 'absolutely' possible and pledged to pause training if needed.
Sam Altman told Fortune this week there will be no OpenAI IPO in 2026, and he tied the delay directly to AI safety rather than to market conditions. "Right now would be an ill-advised moment to go public," Altman said, with "everything happening with safety" as the reason.
"We don't feel pressure on timing," Altman told Fortune, with the company going public "when we're ready." On a 2026 IPO he was direct: "I would say not 2026. We've got a lot of stuff to do." (The Verge)
The same 45-minute interview also covered Altman's view on building AI beyond human control. He said it was "absolutely" possible, and that the industry had a duty to act on that risk before it materialized. "There are risks we should not be able to incur on behalf of humanity," he told Fortune.
In the same conversation, Altman pledged to act on that risk, including potentially "pausing training" if his team concluded the technology was crossing a line they could not yet govern. The IPO delay and the training-pause promise now sit on the same page of the same interview, in front of the same audience.
The interview came shortly after a hacking incident on Hugging Face, an open-source AI platform where developers share and download AI models, which Altman referenced in the same conversation. He is not invoking safety in the abstract. He is citing a specific near-term event that exposed the fragility of the model-distribution pipeline.
By naming a specific, schedulable capital event (the IPO) and tying it to a publicly stated safety readiness criterion, Altman has made OpenAI's safety progress externally auditable, by investors, regulators, and the public, rather than something reviewed only inside OpenAI's own governance. The mechanism is a public-market scrutiny lever that a private-company board cannot apply to itself.
A legitimate counterread is also worth naming: tying IPO timing to safety could be a capital-markets positioning move. It tells prospective public-market investors that the company intends to manage safety as a profit-and-loss-grade risk before they buy in, and it tells critics that the company is taking safety seriously enough to delay liquidity. Both readings are consistent with Altman's exact words. The transcript does not let a reader rule either one in or out.
The Fortune interview was re-reported the same day by The Verge, TechCrunch, Business Insider, and The Guardian Tech. All four outlets trace back to the same underlying conversation, so the analysis rests on Altman's own statements, not on a mosaic of independently sourced claims.
The next milestone to watch is OpenAI's own framing of when safety is ready. Until then, "not 2026" is the answer Altman is willing to put on the record.