Elon Musk's First Day on the Stand: 5 Moments That Defined His OpenAI Trial Testimony
Musk refused yes/no questions, forgot his own testimony, and scolded his defense lawyer. A Verge reporter said they'd never felt more sympathy for Altman.
Five Hours on the Stand, and Elon Musk Made Sam Altman Look Good
The Musk v. OpenAI trial started the week of April 28, 2025, and within five hours of Elon Musk’s testimony, a reporter covering it live for The Verge typed out a sentence that tells you almost everything you need to know about how day one went: “I have never been more sympathetic to Sam Altman in my life.”
That’s the headline. Musk refused yes/no answers, forgot things he’d testified to earlier the same morning, and scolded defense lawyer William Savit in front of the jury. One juror was spotted rubbing her head during a particularly testy exchange. Jury members glanced at each other.
This is a trial Musk chose to bring. He’s the plaintiff. And somehow, by the end of his own testimony, the reporter who’d been watching it unfold in real time had been pushed into sympathy for the man Musk is trying to remove from OpenAI.
Here are the five moments that defined that first day — and what each one actually reveals about where this case is headed.
The Setup: What Musk Is Actually Claiming
Before getting into the testimony itself, it’s worth being precise about what Musk walked into that courtroom to argue.
Other agents start typing. Remy starts asking.
Scoping, trade-offs, edge cases — the real work. Before a line of code.
His core claim is that Sam Altman and Greg Brockman tricked him into donating money to OpenAI, then abandoned the founding mission of developing AI to benefit humanity in favor of profit. He wants Altman and Brockman removed from the company. He also wants OpenAI to stop operating as a public benefit corporation — which is a specific legal structure, not just a vibe.
OpenAI’s defense is more pointed: this lawsuit is a bid to boost Musk’s own competing companies — SpaceX, XAI, and X — which together house Grok, a direct competitor to ChatGPT. The argument is essentially that Musk isn’t a wronged idealist; he’s a competitor using the courts as a business strategy.
Those are the two poles. Everything that happened during testimony has to be read against that backdrop.
Moment 1: The Yes/No Problem
The most consistent thread through the day’s coverage was Musk’s refusal to answer yes/no questions with yes or no.
This sounds like a small thing. It isn’t. In a courtroom, when a lawyer asks a yes/no question and the witness refuses to answer it directly, there are usually two explanations: the honest answer is damaging, or the witness doesn’t trust themselves to stop talking. Either way, it’s a problem.
For Musk, this pattern repeated throughout the day. Defense lawyer William Savit would ask a direct question. Musk would respond with something longer, more qualified, or simply not responsive. The Verge’s live updates, which were posting roughly every 15 minutes, captured this dynamic repeatedly.
The strategic risk here is real. Jurors aren’t legal experts, but they’re human beings who can tell when someone is dodging. When the person dodging is the plaintiff — the one who brought the case — it raises an obvious question: if your position is so clearly right, why can’t you just say yes?
Moment 2: Forgetting His Own Morning Testimony
By the afternoon, Musk had apparently forgotten things he’d said on the stand earlier that same day.
This is the detail that’s hardest to explain away. Forgetting something from a deposition taken months ago is understandable. Forgetting your own testimony from the morning session of the same day is a different category of problem.
The Verge reporter noted this specifically — that Musk “occasionally forgot things he’d testified to in the morning.” In a trial where Musk’s credibility is central to his claims (he’s arguing he was deceived, which requires the jury to believe his account of events), this kind of inconsistency is corrosive.
It also hands the defense exactly what they need. If Musk can’t keep his own story straight across a single day, it becomes much easier to argue that his recollection of conversations and agreements from years ago — the ones at the heart of his fraud claims — might be equally unreliable.
Moment 3: Scolding the Defense Lawyer
At some point during the day, Musk scolded William Savit, the defense lawyer.
The Verge’s reporter called it out directly in their coverage. And it’s worth sitting with what this means in context.
Savit is doing his job: cross-examining a witness, probing for inconsistencies, asking pointed questions. That’s the adversarial process. Witnesses who get visibly frustrated with opposing counsel — especially to the point of scolding them — tend to come across as either thin-skinned or evasive. Sometimes both.
Not a coding agent. A product manager.
Remy doesn't type the next file. Remy runs the project — manages the agents, coordinates the layers, ships the app.
For Musk specifically, this is a pattern that people who follow him closely will recognize. He has a documented tendency to treat challenges to his positions as personal attacks, and to respond with escalation rather than engagement. That works fine on X, where he controls the platform and the context. It does not work well in a federal courtroom, where the rules are different and the audience — twelve jurors — is watching everything.
The woman rubbing her head during the testy exchange is the image that sticks. She’s not a pundit. She’s not a tech journalist with a take. She’s a juror, and she looked exhausted.
Moment 4: The Verge Reporter’s Verdict
The Verge’s real-time update page is worth understanding as a source here. It’s not an opinion column — it’s live courtroom coverage, updating every 15 minutes or so, written by a reporter who is presumably trying to be fair to both sides.
That’s what makes the sentence so striking: “I have never been more sympathetic to Sam Altman in my life.”
This isn’t someone who came in rooting for Altman. This is someone who spent five hours watching Musk testify and came out the other side with a reaction they apparently felt compelled to type out and publish. The phrasing — “I typed the following sentence” — suggests it surprised even the reporter writing it.
The full context matters: “For hours, Musk refused to answer yes or no questions with yes or no. Occasionally forgot things he’d testified to in the morning, and scolded defense lawyer William Savit. I watched a few jury members glance at each other. During one testy exchange, one woman was rubbing her head.”
That’s not a summary of a strong plaintiff’s performance. That’s a summary of a day that went badly for the person who chose to be there.
Moment 5: The Jury’s Body Language
Courtroom observers pay attention to jurors for a reason. Jurors can’t speak, but they communicate constantly — through attention, through restlessness, through the small physical tells that signal whether they’re engaged or checked out.
The Verge reporter noted two specific moments: jury members glancing at each other, and one woman rubbing her head during a testy exchange.
Glancing at each other is interesting. It can mean a lot of things — shared confusion, shared skepticism, shared amusement. But when it happens during testimony from the plaintiff, during a moment of evasion or frustration, it’s rarely a good sign for the plaintiff’s case.
The woman rubbing her head is more legible. That’s the body language of someone who is tired of what they’re watching. Not convinced by it. Not moved by it. Tired of it.
None of this is determinative. Trials are long. Jury impressions shift. There’s more testimony to come from both sides, and OpenAI’s defense hasn’t fully made its case yet. But first impressions in a courtroom matter, and Musk’s first impression appears to have been: difficult, evasive, and occasionally forgetful.
What the Testimony Reveals About the Underlying Case
Step back from the individual moments and a pattern emerges.
Musk’s legal theory requires the jury to believe several things simultaneously: that he was deceived by Altman and Brockman, that his motivations were purely idealistic, and that his current legal action has nothing to do with the fact that his own AI company (XAI) and his own AI product (Grok) are in direct competition with OpenAI.
One coffee. One working app.
You bring the idea. Remy manages the project.
That last part is the hardest sell. OpenAI’s defense is essentially asking the jury to apply Occam’s razor: the simplest explanation for why a billionaire with a competing AI company is suing to remove the leadership of that competitor is competitive interest, not wounded idealism.
Musk’s testimony needed to counter that narrative. It needed him to come across as credible, consistent, and genuinely aggrieved — not as someone who dodges questions, loses track of his own story, and snaps at lawyers.
The broader context here is worth keeping in mind. OpenAI’s own corporate evolution — from nonprofit to capped-profit to its current structure — has been genuinely complicated, and the relationship between Anthropic, OpenAI, and Google’s different strategic bets on AI reflects just how much the landscape has shifted since OpenAI’s founding. Musk’s argument that the company abandoned its mission isn’t inherently absurd. But “not absurd” and “convincing” are different standards, and his testimony apparently didn’t clear the second bar.
The questions about OpenAI’s direction also connect to real debates about model development and pricing. Understanding how token-based pricing works — and how dramatically it’s shifted as competition has intensified — gives some texture to why OpenAI’s commercial pivot matters to the people actually building on these models. The company Musk is suing is the same company that set the pricing benchmarks everyone else is now racing to undercut.
The Deeper Irony
There’s something almost structurally ironic about how this day played out.
Musk built much of his public persona on the idea that he’s a straight talker — someone who says what others won’t, who doesn’t hedge, who operates without the corporate caution that makes most executives sound like they’re reading from a script. That persona is a large part of why he has the following he does.
But in a courtroom, where straight answers are literally what’s being asked for, he apparently couldn’t deliver them. Yes/no questions got non-answers. Morning testimony got forgotten by afternoon. A lawyer doing his job got scolded.
The trial is ongoing, and there’s more to come. OpenAI still has to make its case. Witnesses will be called. Documents will be entered into evidence. The jury hasn’t decided anything yet.
But if you’re trying to understand what happened on day one, the Verge reporter’s sentence is the cleanest summary available: five hours of testimony, and the person who walked out with more public sympathy was the defendant.
What Comes Next
The trial will continue, and the coverage from The Verge’s live update page will keep filling in the details. The structural questions — whether Musk was genuinely defrauded, whether OpenAI’s conversion away from nonprofit status was a betrayal of its founding mission, whether Musk’s competing interests disqualify his claims — will eventually get answered by the jury.
What’s harder to recover from is a first day that handed the other side a narrative. OpenAI’s lawyers don’t need to prove Musk is a bad person. They need to create enough doubt about his credibility and his motives that the jury can’t confidently rule in his favor. Based on the day one coverage, Musk may have done some of that work for them.
- ✕a coding agent
- ✕no-code
- ✕vibe coding
- ✕a faster Cursor
The one that tells the coding agents what to build.
The developments around OpenAI’s model roadmap — including what comes after GPT-5 — will keep moving regardless of how the trial ends. The company isn’t pausing product development while its co-founder tries to remove its CEO in court. That’s probably the most telling thing about where OpenAI actually is right now: the trial is a sideshow to the main event, even if it’s a very loud one.
For builders and engineers trying to make sense of which AI infrastructure to bet on, the trial is mostly noise. The signal is in the models, the pricing, and the APIs. But the human drama of it — a founder suing the company he helped start, making the defendant look sympathetic in the process — is a reminder that the AI industry’s biggest decisions are still being made by people, with all the irrationality that implies.
Tools like MindStudio are built on the assumption that what matters is what you can build, not who’s suing whom — 200+ models, 1,000+ integrations, and a visual builder for chaining agents and workflows, all of it available regardless of how the Musk v. OpenAI case resolves. The underlying infrastructure keeps compounding either way.
And for teams thinking about how to build more durable software on top of that infrastructure, Remy represents a different kind of bet: write your application as an annotated spec, and compile it into a complete TypeScript stack — backend, database, auth, deployment. The spec is the source of truth; the code is derived output. It’s a useful frame for thinking about what “ownership” means when the tools keep changing underneath you.
The trial will produce a verdict eventually. The models will keep shipping in the meantime. That asymmetry is probably the most important thing to understand about this particular week in AI.
