The week leading up to Thanksgiving 2023 didn’t just change the leadership of OpenAI—it exposed, in real time, how fragile trust can be inside a company where decisions move faster than consensus. Sam Altman’s abrupt ouster as CEO became one of the most watched corporate shakeups in AI history, and the explanation offered at the time was both specific in its accusation and vague in its mechanics: the board said Altman was “not consistently candid in his communications with the board.”
For months afterward, the public filled in the gaps with speculation. Who knew what, when? What exactly was “candid” versus “inconsistent”? Was this primarily about governance and oversight, or did it reflect deeper disagreements about direction, risk, and control? The problem with corporate narratives is that they often arrive after the fact—sanitized, compressed, and designed to end questions rather than answer them.
Now, through witness testimony and trial exhibits discussed in Musk v. Altman, those missing details are beginning to surface in a more concrete way. And while the case involves multiple parties and claims, a central figure in the newly illuminated timeline is former OpenAI CTO Mira Murati. Her deposition and related evidence don’t simply add color to an already dramatic story; they help clarify how the board’s concerns translated into actions, how internal messaging evolved under pressure, and why the weekend’s sequence mattered as much as the final outcome.
What makes this period especially striking is not only the speed of events, but the way the speed itself appears to have shaped the governance process. In many corporate crises, there’s time for deliberation, for careful drafting, for controlled communication. Here, the record suggests something closer to a feedback loop: leadership and board communications influenced each other hour by hour, and the uncertainty created by shifting messages likely affected how people interpreted subsequent developments. That dynamic—where trust erodes while information changes—can turn a governance dispute into a legitimacy crisis.
To understand why Murati’s testimony is so consequential, it helps to focus on what her role represented inside OpenAI. As CTO, she wasn’t merely a senior executive; she was a technical leader positioned near the center of product and research execution. That matters because governance disputes often get framed as abstract boardroom disagreements. But when a technical executive is pulled into the narrative, the dispute stops being purely procedural. It becomes a question of how leadership decisions affect the organization’s ability to operate, communicate, and execute.
In the weeks before Altman’s removal, OpenAI was already navigating intense external scrutiny and internal pressure. The company’s mission and pace meant that leadership credibility wasn’t just a matter of optics; it was tied to operational continuity. When the board moved to remove Altman, the company didn’t have the luxury of treating the event as a distant corporate adjustment. It landed on teams that had to keep building while leadership status changed around them.
The board’s stated reason—Altman’s lack of consistent candor—sounds like a governance principle. Candor is a standard used in oversight relationships: executives owe boards accurate, timely, and complete information so directors can make informed decisions. But the phrase “not consistently candid” raises immediate questions that the public couldn’t answer at the time. Consistently how? In what channels? About what topics? Was the issue about withholding information, about timing, about interpretation, or about disagreement over what should be disclosed?
The new material discussed in Musk v. Altman doesn’t magically resolve every ambiguity, but it does shift the story from rumor to documented testimony. Instead of relying on secondhand accounts, observers can now look at how witnesses describe the communications environment and how the board’s concerns were understood internally. That’s a meaningful difference. Corporate governance disputes are often fought over narratives—what was said, what was meant, and what was known. Trial records, by their nature, force specificity.
One of the most important themes emerging from the discussion around Murati’s deposition is that the situation appears to have been less like a single decision and more like a sequence of events shaped by communication patterns. The board’s early messaging, described as broad or vague, didn’t provide the kind of operational clarity that employees and leadership need during a crisis. When communication is vague, people fill in the blanks. And when the blanks are filled with fear, loyalty, or competing interpretations, the organization’s internal alignment can fracture quickly.
This is where the “hour-to-hour” quality of the weekend becomes more than a dramatic detail. If decisions and messaging shifted rapidly, then the organization’s understanding of what was happening would also shift rapidly. That can create a compounding effect: each new statement changes the meaning of the previous one, and each reinterpretation affects how people respond. In governance terms, that’s dangerous because oversight depends on stable information flows. In organizational terms, it’s destabilizing because it turns leadership into a moving target.
Murati’s involvement in the record matters because she sits at the intersection of technical execution and executive leadership. When a technical leader is asked to testify about internal communications, it signals that the dispute wasn’t confined to board minutes. It touched the broader leadership ecosystem—how executives understood the board’s position, how they reacted, and how they communicated among themselves and with teams.
There’s also a subtler point: candor disputes are rarely only about facts. They’re also about relationships and expectations. A board may believe it is owed a certain standard of disclosure. An executive may believe they are providing sufficient information, or that the board is misunderstanding the context. Even if both sides are describing the same events, they may disagree about what level of detail was necessary, what was urgent, and what could wait.
That’s why the phrase “not consistently candid” is so potent. It implies a pattern, not a one-off mistake. But patterns are hard to prove without a timeline. The value of deposition testimony and trial exhibits is that they can anchor the narrative to dates, messages, and described interactions. In other words, they can show whether the board’s claim reflects a sustained breakdown in communication—or whether it reflects a disagreement about interpretation that later hardened into a governance justification.
Another key element in the public’s understanding of the Altman ouster has been the role of Elon Musk, and the broader legal framing of Musk v. Altman. While the case is not simply a retelling of the Thanksgiving weekend, it functions as a mechanism for testing claims against evidence. That means the record is likely to include not only what people believed, but what they said and did. For readers, that matters because it reduces the space for purely speculative storytelling.
Still, it’s important to avoid the trap of thinking that court records automatically produce a clean, definitive truth. Depositions and exhibits are evidence, not omniscience. Witness memory can be imperfect. Documents can be incomplete. And legal strategies can shape what gets emphasized. But compared to the earlier public narrative—built largely on partial statements and anonymous sourcing—the shift toward documented testimony is a meaningful improvement in clarity.
So what does this new clarity suggest about how OpenAI’s governance crisis unfolded?
First, it suggests that the board’s concerns about candor were not merely rhetorical. They were treated as a governance issue serious enough to justify removal. That alone indicates that the board believed the problem undermined its ability to oversee the company effectively. In a fast-moving AI environment, oversight failures can become existential risks. If directors believe they are being misled or not fully informed, they may conclude that continued leadership under the executive is incompatible with their fiduciary responsibilities.
Second, it suggests that internal communications—both within leadership and between leadership and the board—were central to the crisis. The weekend’s drama wasn’t only about who held the title of CEO. It was about who controlled the narrative of what was happening and why. When communications are inconsistent, people interpret silence as concealment and partial updates as strategic maneuvering.
Third, it suggests that the crisis escalated partly because the organization had to operate under uncertainty. Employees and executives can handle disagreement, even intense disagreement, if they understand the boundaries and the process. But when the process itself appears unstable—when decisions seem to change quickly and explanations remain broad—uncertainty becomes a force multiplier. It can accelerate departures, intensify internal factions, and create a sense that normal governance mechanisms are failing.
Murati’s deposition, as discussed in the context of the case, appears to illuminate these dynamics by showing how leadership perceived the board’s actions and how those perceptions evolved. That doesn’t necessarily mean Murati is portrayed as the architect of any outcome. Rather, her testimony provides a vantage point: she is close enough to leadership operations to describe how the crisis felt from inside the executive layer.
There’s also a broader lesson here about corporate governance in technology companies. Boards often claim they want transparency, but transparency is not just about sharing information—it’s about sharing information in a way that supports decision-making. Candor is not a moral category; it’s an operational one. If the board believes it cannot rely on the information it receives, it may treat the relationship as broken. But executives may believe they are providing the right information, just not in the format or timing the board expects.
When those expectations diverge, the dispute can become self-reinforcing. The board demands more clarity; the executive provides what they think is clarity; the board interprets it as insufficient; the executive interprets the board’s demands as politicized or unrealistic. Eventually, the board may decide that the relationship cannot be repaired quickly enough to protect the company.
In OpenAI’s case, the stakes were amplified by the company’s public profile and the speed of AI development. Leadership credibility affects partnerships, hiring, investor confidence, and internal morale. So even if the underlying dispute began as a governance disagreement, the consequences quickly became organizational and reputational.
Another unique aspect of this story is how visible parts of the power struggle became to the public. Many corporate leadership disputes are largely hidden until the outcome is announced. Here, the weekend unfolded in a way that made it feel like a live event. That visibility likely