New testimony from co-founder Ilya Sutskever reveals a culture of manipulation, “disappearing” emails, and the desperate bid to save OpenAI from its own CEO.
- The Secret Dossier: Ilya Sutskever compiled a 52-page memo documenting Sam Altman’s alleged “pattern of lying” and manipulation, which he delivered to the board via disappearing emails to prevent Altman from making the evidence vanish.
- Divide and Conquer: Testimony suggests Altman maintained control by pitting top executives against one another, including Sutskever, current Chief Scientist Jakub Pachocki, and former CTO Mira Murati.
- A Miscalculated Coup: Despite the detailed allegations of psychological manipulation and chaos, Sutskever admitted he drastically underestimated the employee loyalty to Altman, assuming the staff would remain neutral regarding the ouster.
It has been nearly a year since the tech world stood still. In November 2023, Sam Altman, the face of the AI revolution, was unceremoniously fired by the board of OpenAI, only to be reinstated days later following an employee revolt. At the time, the board’s vague explanation—that Altman was “not consistently candid”—fueled endless speculation. What could the CEO of the world’s most important startup possibly be hiding?
Thanks to a marathon 10-hour legal deposition of OpenAI co-founder Ilya Sutskever, conducted as part of Elon Musk’s lawsuit against the company, we finally have an answer. The testimony paints a portrait of a Machiavellian leadership style where information was weaponized, executives were turned against one another, and a 52-page dossier of grievances became the catalyst for the most dramatic boardroom coup in Silicon Valley history.
The “Consistent Pattern” of Deceit
For months leading up to the firing, Ilya Sutskever, then OpenAI’s Chief Scientist, had been quietly chronicling his dissatisfaction. According to the deposition, the independent board members—Adam D’Angelo, Helen Toner, and Tasha McCauley—asked Sutskever to formalize the concerns swirling among the executive team.
Sutskever responded with a 52-page memo, sent via disappearing emails to ensure secrecy. When asked why he didn’t confront Altman directly, Sutskever’s answer was telling: “Because I felt that, had he become aware of these discussions, he would just find a way to make them disappear.”
The memo, parts of which were read during the deposition, was damning. Its opening page alleged that Altman “exhibits a consistent pattern of lying, undermining his execs, and pitting his execs against one another.” Sutskever also submitted a separate memo detailing concerns regarding OpenAI President Greg Brockman, suggesting the rot at the top went beyond just the CEO.
Management by Chaos
The testimony offers a rare glimpse into the internal mechanics of OpenAI, describing an environment rife with manufactured conflict. Sutskever testified that Altman frequently told different executives “conflicting things about the way the company would be run,” effectively paralyzing them or setting them on collision courses.
Sutskever cited specific instances of this “divide and conquer” strategy:
- Sutskever vs. Pachocki: Altman allegedly played Sutskever against Jakub Pachocki (who has since replaced Sutskever as Chief Scientist), undermining Sutskever’s authority by offering contradictory directives.
- The Anthropic Split: When Dario Amodei (now CEO of rival Anthropic) was still at OpenAI, he reportedly demanded to run all research and wanted Greg Brockman fired. Sutskever claimed Altman refused to accept or reject these terms, playing both sides to see which outcome would best serve his own power base.
- Murati vs. The Amodeis: Former CTO Mira Murati provided Sutskever with screenshots and documentation suggesting Altman had pitted her against Daniela Amodei, further fracturing the leadership team.
Perhaps most damaging was the revelation that this behavior wasn’t new. Murati had surfaced claims that Altman’s previous departure from Y Combinator was due to “similar behaviors,” alleging he created chaos and started unauthorized projects, proving he “was not managing YC well.”
The Board in the Dark
Sutskever’s testimony corroborates public statements made by former board member Helen Toner in May 2024. Toner described a board that was systematically blinded, learning about the launch of ChatGPT on Twitter and being misled about the company’s safety processes. She alleged Altman even concealed his ownership of the OpenAI Startup Fund.
“All four of us who fired him came to the conclusion that we just couldn’t believe things that Sam was telling us,” Toner stated. The board had reportedly been told by executives—likely Sutskever and Murati—that Altman’s leadership amounted to “psychological abuse” and that he was not the right person to lead humanity toward Artificial General Intelligence (AGI).
The Failed Merger and the Exodus
The deposition also sheds light on the frantic days following Altman’s firing. In a twist that could have reshaped the AI landscape, rival company Anthropic approached OpenAI with a merger proposal. Sutskever recalled a call involving Dario and Daniela Amodei, where they expressed excitement about taking over leadership. However, the deal collapsed quickly due to “practical challenges.”
Ultimately, the coup failed. Altman returned, and the “consistent candor” issue was swept away by an internal investigation that concluded Altman and Brockman were the right leaders. OpenAI spokesperson Liz Bourgeois reiterated this stance, stating, “The events of 2023 are behind us.”
The human cost of the conflict was total. Sutskever remained at the company in name only for six months before leaving to found Safe Superintelligence (SSI). Mira Murati, Bob McGrew, and Barret Zoph followed him out the door shortly after.
In a final moment of reflection during the deposition, Sutskever admitted to a significant miscalculation. When asked if he thought the staff would support the board’s decision to remove Altman, he replied, “I had not expected them to cheer. But I [had] not expected them to feel strongly either way.”
He was wrong. The employees threatened to resign, Altman returned, and Sutskever—once the visionary behind OpenAI’s technology—found himself on the outside, his 52-page warning ignored by history, if not by the court record.