OpenAI governance – MISRYOUM reports Greg Brockman’s court testimony about a tense 2017 negotiation with Elon Musk and why he questioned Musk’s control of OpenAI.
Elon Musk’s bid to shape OpenAI may have started with gifts and charm, but Greg Brockman’s court testimony paints a far more volatile picture from a 2017 meeting at Musk’s “haunted” estate.
Brockman. who co-founded OpenAI alongside Ilya Sutskever. described how an in-person discussion about the future of AI and the company’s structure turned sharply tense.. The meeting took place at Musk’s large Hillsborough property. where Misryoum reports Brockman testified that Tesla Model 3 cars were presented as part of the lead-up to negotiations.. In Brockman’s account, it felt like Musk was trying to put the OpenAI founders in his debt.
Insight: This matters in the broader AI story because governance decisions are often treated as technical and organizational matters, yet they can be deeply shaped by personal power dynamics.
According to Brockman’s testimony. Musk wanted control as OpenAI explored the idea of a for-profit arm designed to attract significant investment for compute.. Brockman and Sutskever pushed back. arguing that giving Musk authority would amount to a level of control they saw as risky for the direction of AI development.. After a proposal was rejected, Brockman recalled the atmosphere escalating quickly, including concerns about physical confrontation.
In the same testimony. Brockman also referenced later experiences meant to highlight his skepticism about Musk’s ability to guide an AI organization.. Misryoum reports Brockman described how early chatbot behavior disappointed Musk, and how that reaction affected a researcher’s confidence.. In Brockman’s view. failing to “dream” about the potential of early AI systems was a sign Musk wasn’t suited to lead decisions at OpenAI.
Insight: When early AI products are dismissed too quickly, it can discourage iteration, which is central to turning prototypes into reliable technology.
Brockman further told the court that discussions among OpenAI leadership included the possibility of voting Musk off the nonprofit board during months of ongoing talks about the proposed structure.. Misryoum reports that he and Sutskever ultimately decided removing Musk felt “wrong. ” even as negotiations continued and alternative funding ideas were explored.
The dispute is tied to claims in the Musk v.. Altman case, where Musk argues donations to OpenAI were mishandled, while OpenAI leaders deny wrongdoing.. Misryoum notes that Brockman’s testimony also included references to continued communications through Shivon Zilis. described in his account as a channel between the two sides as later developments unfolded.
Insight: Beyond the courtroom, these accounts underscore how AI companies can be shaped not only by models and money, but by trust, decision rights, and who gets to steer risk when the stakes rise.