Elon Musk returned to the witness stand for a second day in the ongoing legal fight tied to OpenAI, and this time the spotlight wasn’t just on what Musk has said about artificial intelligence over the years—it was on how those statements are being treated as evidence inside a courtroom.
According to reporting from TechCrunch, Musk’s testimony continued as part of his attempt to legally dismantle OpenAI. The case, at its core, is about governance: who controls the direction of a powerful AI organization, what obligations exist when that organization is structured around public-facing promises, and whether the way OpenAI operates matches the commitments that were made when it was formed and later reorganized. But in practice, governance disputes like this often turn into something more granular and human: intent, expectations, and the record of communications that shaped them.
And in Musk’s situation, that record includes a very specific kind of communication—his own tweets.
That detail matters because courts don’t evaluate arguments in the abstract. They evaluate claims against a timeline of actions and statements. When a party’s public messaging is repeatedly cited, it can serve multiple purposes at once: establishing what they believed at the time, showing what they wanted others to believe, and demonstrating whether their later legal positions align with earlier rhetoric. In other words, tweets can become more than commentary; they can become a map of mindset and strategy.
Day two of testimony, as described by TechCrunch, suggests that Musk’s prior statements are still central to what the court is trying to understand. The continued appearance indicates that the legal questions aren’t simply about whether Musk has strong opinions about AI. They’re about whether those opinions connect to specific allegations in the lawsuit—particularly allegations that hinge on how OpenAI’s leadership and structure have evolved, and whether those changes comport with the promises and constraints that were supposed to guide the organization.
To understand why this is such a high-stakes moment, it helps to recognize what makes this case unusual. Many technology lawsuits involve contracts, patents, or straightforward business disputes. This one is different because it’s fundamentally about institutional design and fiduciary-like responsibilities—about whether an organization built around certain ideals has drifted away from them. That kind of dispute tends to produce evidence that looks less like invoices and more like documents of intent: emails, board materials, internal memos, and—when available—public statements that can be used to infer what key actors expected, feared, or prioritized.
Musk’s tweets are a particularly tempting source for that kind of inference. They are frequent, emotionally charged, and often written in a style that compresses complex ideas into sharp claims. They also have a long shelf life. Even if a tweet is posted casually, it can remain discoverable and quotable years later. In a courtroom, that permanence becomes a weapon for whichever side believes the tweet reflects a consistent position—or a shifting one.
The unique twist here is that Musk isn’t merely being questioned about his conduct. He’s being questioned about his own words.
That dynamic can be uncomfortable for any witness, but it’s especially consequential for someone whose public persona is built on rapid-fire commentary. When you testify, you can explain context, nuance, and timing. But you can’t easily erase the fact that your earlier statements exist, and that opposing counsel can read them back with the implication that they reveal something important about what you knew or intended.
In plain terms, what this means is that the court is likely probing the relationship between Musk’s public narrative and his legal narrative. If the lawsuit argues that OpenAI’s governance has failed in ways that violate commitments or expectations, then the question becomes: what did Musk believe those commitments were? What did he think OpenAI was obligated to do? And did his understanding change as events unfolded?
Tweets can help answer those questions because they capture Musk’s framing at particular moments. They can show whether he consistently criticized OpenAI’s direction, whether he praised certain aspects at earlier stages, and whether his rhetoric shifted in response to specific developments. Even if a tweet is not “evidence” in the way a contract clause is, it can still be used to support an argument about credibility and consistency.
There’s another reason tweets are likely to matter: they can establish notice. In many legal contexts, notice is crucial. If someone publicly claims that an organization is acting improperly, that can be used to argue that the claim wasn’t speculative or invented later. It can also be used to argue that the speaker had enough information—or at least enough concern—to make the statement at the time.
Of course, the defense side may argue that tweets are not formal statements of fact, that they are rhetorical, or that they reflect opinion rather than knowledge. But even that argument has to contend with the fact that Musk’s tweets are often presented as confident assertions. When a witness is asked to reconcile those assertions with later legal positions, the courtroom becomes a place where style meets substance.
What makes this second day especially notable is that it signals continuity. If the first day focused on broad background—how Musk became involved, what he alleges, and what he expects from the court—then day two appears to be narrowing in on the record. That narrowing is often where cases become more revealing. Broad testimony can establish themes. Focused testimony can test those themes against specific documents and statements.
TechCrunch’s framing—“can’t escape his own tweets”—captures the emotional reality of the moment, but the deeper significance is procedural. The court is likely working through the evidentiary chain: which statements are relevant, how they relate to the alleged governance failures, and what they suggest about intent and understanding.
This is also where the case intersects with a broader debate about AI governance. Musk is not just a litigant; he is a prominent voice in the AI ecosystem. His public statements influence how investors, policymakers, and technologists interpret the stakes of AI development. When a case involving AI governance reaches the point where a witness’s tweets are scrutinized, it underscores how intertwined public discourse and institutional power have become.
AI governance is often discussed in terms of principles—safety, alignment, transparency, accountability. But in practice, governance disputes are about mechanisms: board composition, decision rights, oversight structures, and the enforcement of obligations. Those mechanisms are rarely debated in the same language as public tweets. Yet the tweets can still matter because they shape expectations and can influence how people interpret the legitimacy of governance decisions.
If Musk’s tweets are being used to illuminate what he believed about OpenAI’s obligations, then the court is effectively asking whether the organization’s governance has lived up to the narrative that surrounded it. That narrative, in turn, has been shaped by Musk’s own public commentary as well as by OpenAI’s messaging.
There’s also a strategic dimension. Musk’s legal effort to dismantle OpenAI is not just about stopping a company; it’s about changing an institution. That kind of effort requires persuasion, and persuasion in court often depends on demonstrating that the plaintiff’s concerns are grounded in a coherent understanding of what went wrong. Tweets can either support that coherence or complicate it, depending on how they align with the allegations.
For example, if Musk’s tweets show a consistent pattern of criticism tied to specific governance concerns, that could strengthen the argument that he wasn’t acting opportunistically. Conversely, if the tweets show shifting positions or rhetorical exaggeration, the defense may use that to undermine credibility. Either way, the court is likely to treat the tweets as more than background noise.
Another angle worth considering is how testimony can reveal the limits of public communication. Tweets are designed for speed and reach. Court testimony is designed for precision and accountability. When a witness is questioned about tweets, the witness must translate informal language into legal meaning. That translation process can expose gaps between what was meant and what was understood.
In many cases, witnesses respond by emphasizing context: what was happening at the time, what they knew, what they were reacting to, and what they intended to communicate. But context is not always enough. Opposing counsel can press for specifics: Did you know X? When did you learn it? What did you mean by Y? Why did you phrase it that way? Were you speaking as a factual reporter or as a critic? Did you intend to influence governance outcomes?
Those questions can turn a seemingly simple tweet into a multi-layered inquiry about knowledge, intent, and responsibility.
The most interesting part of day two, based on the reporting, is that the court appears to be continuing down this path rather than moving on quickly. That suggests the tweets are not a side issue. They are part of the core evidentiary work.
In practical terms, the court may be using the tweets to establish one or more of the following:
First, a timeline of Musk’s concerns. If the tweets show that Musk raised governance issues at particular points, the court can compare those points to organizational changes at OpenAI. That comparison can help determine whether Musk’s concerns were reactive or anticipatory, and whether they align with the alleged misconduct or mismanagement.
Second, a record of Musk’s understanding of OpenAI’s structure. Governance disputes often depend on what the plaintiff believed the organization was supposed to do. Tweets can reveal whether Musk thought OpenAI was bound by certain constraints, and whether he believed those constraints were being violated.
Third, a credibility assessment. Courts weigh credibility constantly. If a witness’s public statements appear inconsistent with their testimony, that inconsistency can become a focal point. If the statements are consistent, they can reinforce the witness’s narrative.
Fourth, a question of intent. While intent is notoriously difficult to prove, it can be inferred from patterns of communication. Tweets can show what a witness prioritized—safety, control, accountability, or competition—and whether those priorities match the legal theory.
None of these outcomes are guaranteed. But the fact that day two continues to revolve around tweets implies that the court sees them as relevant to the legal questions.
For observers, this is also a reminder that AI governance is not only about technical systems. It’s about institutions and the people who lead them, and those people communicate constantly—through press