This video breaks down a recently released 62-page deposition of OpenAI co-founder Ilia Sutskever, conducted as part of the lawsuit filed by Elon Musk against Sam Altman. The deposition provides an insider's account of the dramatic events leading to Sam Altman's temporary ousting. Key revelations include Sutskever’s creation of a 52-page memo, compiled with evidence from executives like Mira Murati, which accused Altman of a "consistent pattern of lying" and undermining his team. Sutskever admits the firing was "rushed" due to an inexperienced board and expresses his surprise at the fierce employee loyalty to Altman. The deposition also uncovers a previously unknown proposal from competitor Anthropic to merge with and take over OpenAI's leadership during the turmoil, a move Sutskever strongly opposed. The document paints a detailed picture of the internal power struggles, long-standing grievances, and strategic miscalculations that defined the OpenAI leadership crisis.
The OpenAI Coup: Revelations from Ilia Sutskever's Deposition
0:55 Context of the Lawsuit: The deposition is part of Elon Musk's lawsuit against Sam Altman, which alleges that Altman improperly converted the non-profit OpenAI into a for-profit entity, betraying its founding mission.
2:12 The 52-Page Memo: Ilia Sutskever, at the request of independent board members, compiled a 52-page memo to justify firing Sam Altman. It accused Altman of a "consistent pattern of lying, undermining his execs and pitting his execs against one another."
3:21 Fear of Sam Altman: Sutskever sent the memo using a disappearing email service because he feared Altman, whom he described as a political "killer," would find a way to "make the discussions disappear" if he found out.
4:17 The Anti-Altman Alliance: The memo was created in collaboration with board members Adam D'Angelo, Helen Toner, and Tasha McCauley. OpenAI's CTO, Mira Murati, was a key source, providing screenshots as evidence of Altman's behavior.
11:11 A "Rushed" and Failed Firing: Sutskever attributes the failure of the coup to a "rushed" process handled by a board that was "inexperienced in board matters."
14:10 Board Member's Extreme Stance: Board member Helen Toner reportedly told executives that allowing OpenAI to be destroyed in Altman's absence would be "consistent with the mission."
15:52 Unverified Allegations: The deposition reveals that Mira Murati was the source for claims that Altman had been pushed out of Y Combinator and Greg Brockman had been fired from Stripe. Sutskever admitted he never verified this information.
18:25 Internal Leadership Conflict: A major point of internal friction involved Dario Amodei (now CEO of competitor Anthropic), who wanted to run all of research at OpenAI and have Greg Brockman fired. Sutskever faulted Altman for failing to make a decisive choice.
20:28 Anthropic's Takeover Attempt: A bombshell revelation is that during the chaos of Altman's firing, Anthropic proposed a merger with OpenAI where they would take over the company's leadership. The board was receptive, but Sutskever was "very unhappy" and opposed it.
22:22 A Long-Simmering Plan: Sutskever reveals that he had been considering proposing Altman's removal for "at least a year," but didn't believe it was feasible until the board's composition changed.
22:53 Misjudging Employee Loyalty: Sutskever was "astounded" by the overwhelming employee support for Altman. He had expected employees to be neutral rather than threatening to quit en masse.
24:59 Sutskever's Continued Stake: Despite leaving to start his own company (Safe Super Intelligence), Sutskever still has a financial stake in OpenAI which has increased in value. He also believes OpenAI is paying his legal fees for the deposition.