New reporting describes OpenAI’s November 2023 crisis as more than a sudden boardroom shock. According to a Wall Street Journal report based on the upcoming book The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future by reporter Keach Hagey, tensions over AI safety, company direction and Sam Altman’s management had been building inside the organization before his brief dismissal.
The account centers in part on a private dinner in Los Angeles, where investor Peter Thiel reportedly warned Altman about the influence of the Effective Altruism movement inside OpenAI. That warning, and the events that followed, help explain why the dispute around Altman became one of the most closely watched leadership conflicts in artificial intelligence.
Thiel’s warning pointed to a deeper AI safety split
At the private dinner, Thiel reportedly challenged Altman over what he saw as a growing internal divide. The issue was not only whether OpenAI should pursue rapid product development, but how much weight the company should give to people focused on the existential risks of artificial intelligence.
Thiel specifically raised concerns about the influence of Effective Altruism, or EA, within the company. He told Altman: "You don’t understand how Eliezer has programmed half the people in your company to believe in that stuff." The remark referred to AI researcher Eliezer Yudkowsky, whom Thiel had previously funded.
Altman reportedly rejected the idea that "AI safety people" could take over OpenAI. He pointed to Elon Musk's departure from OpenAI in 2018. Both Musk and Yudkowsky had repeatedly warned about AI's existential risks, and the source article notes that related safety concerns later led seven OpenAI employees to leave and establish Anthropic in 2021.
That sequence matters because it shows that the OpenAI conflict was not simply personal. It was also about how the company should balance AI safety concerns with its commercial direction. The Wall Street Journal account presents that tension as already visible before the board acted against Altman.
Safety leaders later left OpenAI
The divide became easier to see in 2024, when several key AI safety figures departed OpenAI. The source article names Chief Scientist Ilya Sutskever and Jan Leike, who had led the "Superalignment" team.
Leike later criticized OpenAI’s approach to safety in public, saying his team had to fight for computing resources. That detail fits the broader pattern described in the reporting: people working on safety inside OpenAI believed their priorities were not being supported in the way they expected.
The report also says similar issues appeared after the release of GPT-4o in May 2024, when concerns emerged about rushed safety testing. The source does not present these events as isolated complaints. Instead, it connects them to the same pressure point that appeared in the dinner conversation between Thiel and Altman: whether safety concerns had enough real force inside the company.
For readers trying to understand OpenAI’s crisis, that distinction is important. A company can say AI safety matters while still facing internal disagreement over resources, protocols and timing. The reporting suggests that, at OpenAI, those disagreements reached the highest levels of leadership.
Management concerns became part of the board’s case
The Wall Street Journal report also describes months of work by CTO Mira Murati and Sutskever to gather evidence about Altman’s management practices. According to the account, they documented examples of Altman playing employees against each other and bypassing safety protocols.
The report says Altman gave incorrect information to the board during the GPT-4-Turbo launch and was not transparent about his private OpenAI Start-up Fund structure. These claims added another layer to the AI safety conflict. The board was not only weighing abstract questions about risk; it was also evaluating whether Altman’s leadership style created problems for oversight.
Murati’s role is a major part of the new account. When she tried to raise concerns directly with Altman, he reportedly brought HR representatives to her one-on-one meetings for several weeks. Murati then decided not to share her concerns with the board.
That episode shows how strained the internal process had become. The concerns were serious enough to be gathered and documented, but sensitive enough that the people involved appeared to worry about how they could be raised.
The removal moved quickly, then reversed
According to the Wall Street Journal, four board members, including Sutskever, decided through confidential video calls to remove Altman. Sutskever had prepared two extensive documents detailing alleged misconduct by Altman.
The board also decided to remove Greg Brockman. The report says that decision followed Murati’s claim that Brockman had undermined her authority by going around her to communicate directly with Altman.
When the board announced Altman’s removal, it kept the public explanation limited in order to protect its sources. The explanation said only that Altman had not been "consistently candid." But the move quickly ran into a much larger force inside the company: the possibility of a mass employee exodus.
Altman and Brockman were reinstated. Murati and Sutskever signed the letter supporting their return, even though the report describes both as having played earlier roles in the events leading to Altman’s dismissal. Sutskever was reportedly surprised by staff support for Altman because he had expected employees to welcome the change.
The Wall Street Journal findings also conflict with earlier statements, especially Murati’s claim that she was not involved in Altman’s removal. The source article says the systematic gathering of evidence suggests she played a central role. Murati and Sutskever have since left to establish their own AI companies.
What the OpenAI crisis reveals
The reporting adds weight to earlier accounts that AI safety concerns were central to Altman’s brief removal. It also shows that the conflict was not limited to a single decision or a single launch. It involved board trust, internal safety protocols, computing resources, leadership authority and OpenAI’s commercial direction.
For the wider AI industry, the lesson is straightforward: safety debates can become governance debates when the people responsible for risk believe they lack authority, resources or reliable information. At OpenAI, those questions reached the board, reshaped senior leadership and contributed to significant changes in the organization’s approach to AI safety and development.