According to The Verge, new legal testimony from OpenAI co-founder Ilya Sutskever reveals shocking details about Sam Altman’s leadership style that led to his brief November 2023 ouster. During a nearly 10-hour deposition on October 1st as part of Elon Musk’s lawsuit against OpenAI, Sutskever testified that Altman exhibited “a consistent pattern of lying” and pitted high-ranking executives against each other. The chief scientist compiled a 52-page memo with screenshots and documentation that he sent to board members Adam D’Angelo, Helen Toner, and Tasha McCauley, alleging Altman told people “what they wanted to hear” about company plans. Sutskever said he’d been waiting to propose Altman’s removal for “at least a year” before it actually happened, and didn’t share his concerns with Altman directly because he feared Altman would “make them disappear.” The board ultimately reinstated Altman after hundreds of employees threatened to resign, but the new testimony provides the clearest picture yet of what triggered one of tech’s most dramatic leadership crises.
Sponsored content — provided for informational and promotional purposes.
The manipulation playbook
Here’s the thing about Sutskever’s testimony – it paints Altman as almost comically manipulative. We’re talking about a leader who allegedly told both Sutskever and Jakub Pachocki (now OpenAI’s chief scientist) “conflicting things about the way the company would be run,” deliberately setting them against each other. But it gets even more calculated. When former OpenAI research executive Dario Amodei wanted to run “all of research” and have Greg Brockman fired, Sutskever claims Altman refused to take a firm position, essentially playing both sides to see which outcome would benefit him more. Basically, this wasn’t just poor management – it sounds like a deliberate strategy of divide and conquer.
This wasn’t isolated behavior
And the pattern apparently extended beyond just Sutskever. The deposition reveals that former CTO Mira Murati provided Sutskever with screenshots and documentation, including claims that Altman pitted her against Daniela Amodei. Even more damning? Murati allegedly surfaced that Altman left Y Combinator for “similar behaviors” – creating chaos, starting too many projects, and pitting people against each other. When you look at the broader picture, it’s hard to ignore that this might be Altman’s modus operandi rather than a one-time issue. The fact that multiple executives across different companies reported similar experiences suggests something systemic.
The board’s impossible position
Former board member Helen Toner’s earlier comments on a podcast add crucial context here. She said executives used the phrase “psychological abuse” and told the board they “had no belief that he could or would change.” Think about that for a second – we’re not talking about minor personality conflicts, but something severe enough that multiple senior leaders felt compelled to document everything. Toner also revealed that Altman allegedly concealed his ownership of the OpenAI startup fund and that the board learned about ChatGPT’s release on Twitter. When you’re dealing with technology this powerful, that level of opacity from your CEO is genuinely terrifying.
The messy aftermath
So what happened after the failed ouster? Well, the exodus tells its own story. Sutskever technically stayed for six months but then left to start rival company Safe Superintelligence. Murati eventually departed too, launching Thinking Machines Lab. On the same day Murati left, chief research officer Bob McGrew and VP of research Barret Zoph also exited. McGrew was specifically named in Sutskever’s memo as someone the board should speak to about Altman’s behavior. The deposition transcript available through court records also reveals that during the chaos, Anthropic actually proposed merging with OpenAI and taking over leadership, with both Amodei siblings on the call. Toner confirmed this on X, noting it was only briefly considered.
What this really means
Look, the most concerning part isn’t just the interpersonal drama – it’s what this reveals about oversight at one of the world’s most powerful AI companies. When your board can’t trust what the CEO is telling them about safety processes, financial interests, or even major product launches, you’ve got a fundamental governance problem. And Sutskever’s testimony, while detailed, still leaves huge gaps – multiple parts are redacted or missing from the record. The fact that he hasn’t spoken to Altman in 10-12 months or Brockman in 15 months, yet believes OpenAI is probably paying his legal fees, adds another layer of weirdness to this whole situation. Ultimately, this deposition gives us our clearest window yet into why four board members felt they had to fire their CEO, even if it nearly destroyed the company in the process.
