The trial depositions are revealing that OpenAI's 2024 leadership crisis was less corporate governance, more group chat disaster.
The Summary
- New testimony in the Musk v. Altman trial shows OpenAI picked Mira Murati as interim CEO during Sam Altman's ousting via hasty video calls while Altman texted with co-founder about succession
- The 2024 "Blip" when Altman was briefly removed now looks even more improvised than initially reported
- OpenAI's succession planning contrasts sharply with how most billion-dollar companies handle leadership transitions
The Signal
The depositions coming out of Elon Musk's lawsuit against Sam Altman are doing what discovery does best: turning corporate mythology into awkward screenshots. The picture emerging from the trial shows that when OpenAI's board briefly fired Altman in November 2024, the company running the most important AI lab on Earth picked its next CEO the way college roommates pick a restaurant.
Mira Murati became interim CEO not through careful succession planning or board deliberation, but through a series of video calls happening simultaneously with text messages between Altman and other OpenAI leadership. The timeline suggests decisions were being made in real-time, with minimal process, while the AI industry held its breath.
"Sometimes companies pick CEOs based on carefully laid succession plans. Other times, apparently, they pick them on video calls while the current CEO is texting about who the new CEO even is."
This matters because OpenAI isn't just another startup. When you're training frontier models that could reshape every knowledge job on Earth, governance becomes infrastructure. The chaotic nature of the Blip reveals something uncomfortable about how Web4 is being built:
- The companies creating autonomous agents don't have autonomous succession plans
- Multi-billion dollar valuations sitting on governance processes that wouldn't pass a Series A audit
- Critical decisions about AI leadership made with less process than most companies use to approve a marketing campaign
Microsoft had invested billions at this point. ChatGPT had over 100 million users. And the leadership transition looked like a Discord server deciding who gets admin rights.
The Implication
If you're building on OpenAI's APIs or betting your product roadmap on their models, this trial is required reading. The depositions aren't just legal theater. They're a stress test of the governance structures around the tools that are supposed to automate everything else.
Watch what happens when Anthropic, Google, or Meta face similar internal conflicts. The companies building AI agents need better processes than the agents they're shipping. Until then, Web4 infrastructure sits on surprisingly human foundations.