Sam Altman has taken a 2026 OpenAI IPO off the table, saying the company’s safety obligations outweigh any urgency to tap public markets.

In a wide-ranging interview with Fortune, the OpenAI chief said it was “absolutely” possible to build an AI system beyond human control. He said the company would act to stop that outcome — including by pausing training if necessary — because “there are risks we should not be able to incur on behalf of humanity.”¹

That concern shaped his timeline. “We’re not rushing into an IPO,” Altman said, calling the current moment — “given everything happening with safety” — an “ill-advised” time to go public. “I would say not 2026. We’ve got a lot of stuff to do.”²

The argument is not merely about timing. Altman said OpenAI must be able to make choices that are “not obviously in the interest of our business and our shareholders,” a formulation that casts private ownership as a safeguard for hard, potentially costly alignment decisions.³

The announcement arrived as the broader AI industry confronts sharper warnings about runaway systems. Altman’s rival, Anthropic chief executive Dario Amodei, has called for slowing AI development; Altman publicly endorsed the underlying principle, saying the frontier must be paced. The two executives differ in corporate posture, but share the premise that the race cannot be allowed to outrun control.³

Not everyone accepts that tension at face value. David Sacks, reposting criticism aimed at Anthropic, argued that firms asking public investors to support valuations in the trillions cannot simultaneously insist AI is dangerous without answering whether they should be permitted to list at all.David Sacks (DavidSacks)⁴

For Altman, the immediate answer is clear: investors will have to wait. “We cannot take actions that would risk losing control of the future to AI,” he said.³