Trial testimony reveals OpenAI governance delays and decision timing

The OpenAI trial is really about control — but the useful new detail is timing. Not just who wanted OpenAI to make money, but when that became a real decision, who had to approve it, and what promises were actually firm versus half-formed. That is why Greg Brockman’s testimony mattered this week. He was trying to narrow the case from a grand betrayal story into something more mundane and more legally defensible — a messy startup arguing about structure in real time. (cnbc.com) ### What changed this week? On Tuesday, May 5, Brockman wrapped up two days on the stand in federal court in Oakland. He pushed back on Musk’s core claim that OpenAI’s leaders promised to keep the organization nonprofit and then broke that promise. Brockman said he never made that commitment to Musk and never heard anyone else make it in a binding way. (cnbc.com) ### Why were the diary entries such a big deal? Musk’s lawyers used Brockman’s old journal entries to argue that OpenAI’s leadership privately cared about wealth and power long before the company’s later restructuring. That is vivid courtroom material because it gives the jury a clean story — the public mission said one thing, the private thoughts s(cnbc.com)ual promise. Brockman’s defense was basically that the entries captured personal ambition and confusion, not the organization’s final commitments. (arstechnica.com) ### Why does timing matter so much? Because Musk’s case gets much stronger if he can show the mission shift happened early and was concealed. It gets weaker if OpenAI can show that commercial options were discussed openly, evolved over time, and stayed under nonprofit control as the costs of building advanced A(arstechnica.com)tructural changes. (cnbc.com) ### Was Musk always against going commercial? Turns out, that is one of the biggest pressure points in the case. Brockman and OpenAI are trying to show Musk was not objecting to commercialization in principle — he was objecting to losing influence over how it happened. The New York Times report on Tuesday framed Brockman’s testimony this way: Musk h(cnbc.com) lot harder to sustain. (nytimes.com) ### What else came out in testimony? Brockman also said Musk had OpenAI employees do secret work for Tesla on self-driving in 2017. That detail does two things at once. It undercuts the idea that Musk was only defending a pure charitable mission, and it reminds the court that OpenAI’s early governance was loose enough for overlapping loyalties and informal power plays. (cnbc.com) ### Who actually decides this case? Not the jury, at least not in the final sense. A nine-person jury is hearing the liability phase, but its role is advisory. Judge Yvonne Gonzalez Rogers makes the final call, and she would also decide remedies if Musk wins on liability. That means the emotional force of the diary entries matters, but the legal wei(cnbc.com)tually breached a charitable trust. (cnbc.com) ### So what is the real lesson here? The courtroom drama makes this look like a personality war. But the deeper issue is governance lag. OpenAI grew into a company handling enormous capital needs and strategic power before its internal decision rules were clear enough to survive a breakup among founders. When approvals, fallback paths, and evidence standards stay fuzzy, old emails and journals start doing work they were never meant to do. (cnbc.com) ### Bottom line This week’s testimony did not settle whether OpenAI betrayed its mission. But it did sharpen the real question — not whether people around OpenAI talked about money, but when talk hardened into an obligation-breaking decision. (cnbc.com)