Sam Altman rules out an OpenAI IPO in 2026, calling AI extinction risk 'unacceptable' -- while Anthropic pushes ahead on a $2 trillion listing

OpenAI will not go public in 2026, Sam Altman told Fortune in an interview published September 12 -- and the reason he gave wasn't market timing or unfinished corporate restructuring. "Given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that," he said, pushing a listing OpenAI had confidentially filed for in June toward 2027 or later.

Pressed on the specific risk, Altman didn't offer his own number for the odds AI could cause human extinction -- he said he didn't know how such an estimate could even be made. But he was explicit about the threshold that matters:

“Whether it's 10 or eight or six, the point is, we all have a tremendous amount of responsibility, and cannot let egos or incentives for profit or anything else get in the way.” — Sam Altman, to Fortune, September 12

Elsewhere in the same interview, Altman went further than the extinction-risk framing alone: asked whether AI could end up beyond human control, he said that was "absolutely" possible, and that current safety standards are "not at a place" to support pushing capabilities further without addressing it first. That's an unusually blunt admission for a CEO to make about his own company's core product, days after Anthropic's Dario Amodei published an essay calling on AI companies to slow down, citing OpenAI's own coding agents breaching Hugging Face in July as the incident that changed his calculus. Altman had replied to that essay on X saying pacing "has been a primary topic of discussions" at OpenAI -- this interview is the first time that discussion has been tied to a specific corporate decision.

The timing shift is real, but it didn't start this week. OpenAI confidentially filed for an IPO in May, confirmed publicly on June 8, targeting a September debut at a reported $730 billion to $850 billion valuation, with Altman said to treat anything below $1 trillion as a "non-starter." (OpenAI's CFO, Sarah Friar, told employees in August the company expected to go public "in 2027 or sooner" -- weeks before Altman's Fortune interview tied the delay specifically to extinction risk.) That detail matters for how much weight the safety framing can actually carry: the later timeline was already set before this week's interview gave it a public reason.

None of that makes Altman's stated reason fake. It just means it isn't the only plausible one.

Anthropic, meanwhile, is reportedly still moving toward the opposite outcome at almost the same moment: a $2 trillion initial public offering in October, which would be the largest IPO ever. That's the same company whose CEO spent this week arguing the industry needs to pace itself -- worth being precise about what his essay actually committed Anthropic to, which was outside evaluators with employee-level access, not a pause on its own public listing.

What would actually settle which explanation is doing the work is something neither company has offered: a named safety milestone that, once cleared, ends the wait. Altman's interview describes urgency and responsibility in general terms; it names no specific evaluator, benchmark, or date that would mark "the current safety climate" as improved enough to go public after all. Jacob Coxon's resignation from Anthropic last week over the same extinction-risk concerns raised the same underlying question industry-wide: a statement of concern is easy to make; a specific, checkable commitment is the harder thing to point to.