The video reveals that Sam Altman’s sudden firing from OpenAI was a year-long planned coup led by Ilya Sutskever and Meera Miati, involving internal memos accusing Altman of dishonesty and undermining colleagues, amidst board tensions and rumors of a merger with Anthropic. It also highlights complex power struggles, questionable board conduct, and ongoing legal battles, while noting Sutskever’s departure to start a new AI venture and the unclear funding of his legal fees.
The video discusses the recent public release of Ilya Sutskever’s deposition in the ongoing legal battle between Elon Musk and Sam Altman, focusing on the dramatic events surrounding Altman’s brief firing from OpenAI nearly two years ago. The firing was sudden and confusing, with Altman being informed only minutes before his removal, and the official reason given was his inconsistent candor in communications. Emmett Shear, founder of Justin TV (which became Twitch), temporarily took over as CEO. There were also rumors of a potential merger between OpenAI and Anthropic, with Anthropic possibly taking over, adding further complexity to the situation.
Ilya Sutskever revealed in his deposition that the plan to remove Sam Altman had been brewing for over a year, with him and Meera Miati plotting behind the scenes. Dario Amodei, a former OpenAI researcher who later founded Anthropic, wanted Greg Brockman fired and himself in charge of research. Sutskever prepared a 52-page memo outlining reasons to remove Altman, citing a consistent pattern of lying, undermining executives, and creating internal conflicts. He also drafted a similar memo about Greg Brockman. These memos were sent only to independent board members, excluding Altman, to prevent the information from being suppressed.
The deposition also revealed tensions within the board, particularly involving Helen Toner, who was criticized for publicly praising Anthropic and criticizing OpenAI while serving on OpenAI’s board. Sutskever found her actions inappropriate and even discussed the possibility of asking her to leave the board. Toner reportedly suggested that allowing OpenAI to be destroyed could be consistent with its mission, a statement that shocked others. The board was divided, with some members supporting a merger with Anthropic shortly after Altman and Brockman’s removal, though practical obstacles prevented the deal.
Much of the information in the memos and deposition was based on secondhand accounts, primarily from Meera Miati, without direct verification from the individuals involved. There were also discussions about internal safety procedures for AI model releases, with some uncertainty about whether proper protocols were followed for GPT-4 Turbo. The deposition highlighted the complex interpersonal dynamics and power struggles within OpenAI’s leadership, including accusations of lying, undermining colleagues, and failing to take firm stances on critical issues.
Finally, the deposition touched on Sutskever’s departure from OpenAI to start a new company focused on a different approach to AI research. There was also intrigue around who was paying Sutskever’s legal fees, with indications that OpenAI might be covering them despite no clear communication about this arrangement. The video concludes by reflecting on the murky details of the coup against Altman, the ongoing legal disputes, and the complicated relationships among key figures in the AI industry, inviting viewers to consider their own opinions on the matter.