Altman Says Rivals Left OpenAI Because the 'Stakes Feel So High' They Trust Only Themselves

OpenAI CEO Sam Altman has a theory about why so many of the people who helped build modern AI ended up building rival labs: they don't trust anyone else to do it.

Altman made the case on "Figuring Out with Raj Shamani," a roughly 70-minute interview released Nov. 3, 2026. Host Raj Shamani asked why the AI industry has so much friction and why so many people leave OpenAI to start competing projects.

"I think there is something that happens in many people's brains where they only trust themselves to build this because the stakes feel so high," Altman said.

Shamani pressed on whether the split is about how AI should be built or about fear of too much power in one person's hands. Altman said the two are tied together. "I think it's somehow this approach is bad and so I can't let power get concentrated in his hands. Something like that," he said. "It's approach and trust."

Who left, and who has spoken

The Times of India's account of the exchange ties the comments to Elon Musk, Ilya Sutskever, Mira Murati and Dario Amodei. All four left OpenAI and are associated with competing efforts.

None of them is quoted here. Altman is describing their motives secondhand. The departures happened at different times, in different roles, for different stated reasons, and a single quote does not establish a shared motive.

The framing also cuts both ways. Altman casts the exits as a failure of trust by those who left. Anyone who left over fear of concentrated power would put it differently.

The "villain" question

Shamani also noted that Altman helped spark the current AI boom, holds no direct equity in OpenAI, and is still cast as a villain in U.S. and Indian coverage. Altman put that down to American society's wariness of power, especially the vast control wielded by tech giants.

"On one hand, maintaining such oversight and vigilance is a good thing," he said, "but sometimes it does feel like it crosses the line." He called the public's reading of his equity decision one of the biggest misunderstandings about him.

That skepticism is not exactly irrational. A company that says it is building technology of civilizational importance is asking for a lot of trust, and OpenAI's own CEO just described trust as the thing that splits the industry.

What the company has put on paper

OpenAI updated its policy on employee concerns on Oct. 30. Under the policy announcement, staff can take AI safety concerns, or violations of the company's AI safety policies, directly to the safety committee of the OpenAI Foundation board. The policy also bars harassment or retaliation against anyone who raises concerns.

OpenAI's published structure puts the OpenAI Foundation in control of the for-profit OpenAI Group PBC. The Foundation holds special voting and governance rights and can appoint and replace Group directors.

That is the formal design. OpenAI has not disclosed how often the safety channel has been used or how any concerns were resolved.

Altman also said in the interview that a leader should avoid surrounding himself with people who only agree, and that overriding everyone's advice should be rare.

Claims that AI has passed the scientists

Separately, at a conference in Abu Dhabi, Altman told the editor-in-chief of the UAE state-owned paper The National that people should stay in control of the future. "At the start of this year, AI was mostly not as smart as our smartest scientists," he said. "And now it's mostly smarter than them."

Futurism, which has been sharply critical of the claim, called it one that "needs proof." Its account says OpenAI announced a solution to the Navier-Stokes equations, one of the Millennium Prize Problems, and then released hundreds of studies documenting purported solutions to other math problems. Futurism reports that a mathematician accused OpenAI of unfairly using his research and that prominent mathematicians objected to being treated as unpaid proofreaders.

Futurism also says the proofs appear technically correct but are written in language it calls "borderline incomprehensible," and that unpacking them could take researchers years. That is Futurism's assessment. OpenAI has not published an independent verification in the material gathered for this story.

The RAM remarks

Altman has also been cleaning up comments on the Cheeky Pint podcast, where he said, "It's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer." He suggested that people use that hardware only about 5% of the time.

Users pushed back that their personal memory use is tiny next to data centers, which many blame for soaring prices on computers and game consoles. Altman briefly changed his X handle to "Ram Altman." He then posted that he wants the world to have "tons of memory" and that people should put as much as they want in their computers. He said the remarks were taken out of context.

OpenAI did not immediately respond to CNET's request for comment.

What comes next

The first test is the Oct. 30 safety policy. Whether employees use the direct line to the Foundation's safety committee, and what that committee does with what it hears, will say more about OpenAI's internal trust than any interview. The second is the math. Until outside mathematicians finish checking the proofs OpenAI released, Altman's claim that AI now outthinks the best scientists rests on his word and on work that is still being reviewed.