OpenAI Coup Night Text Messages Exposed: Altman Asks "Microsoft, Please Acquire Me"

In that chaotic weekend of November 2023, Sam Altman was suddenly fired by the board. Two and a half years later, his text message exchanges with Mira Murati were publicly revealed in court for the first time, like a live thriller, reconstructing every dramatic moment of Silicon Valley’s most theatrical power struggle frame by frame.

In May 2026, the second week of the OpenAI lawsuit filed by Elon Musk began. In court, an unreleased set of text messages was made public—on the night of the November 2023 coup, the real-time conversation between the recently dismissed Sam Altman and former OpenAI CTO Mira Murati.

These messages are not retrospective recollections, nor secondhand accounts. They are every second of that moment. After reading these exchanges, one would realize: that coup was far more chaotic, absurd, and cold-blooded than media depicted.

Time: Friday, November 17th, afternoon, Altman was just told over the phone that he was fired.

Altman clearly hadn’t processed it yet. He texted Murati:

Altman: Can you tell me if the situation is good or bad? Satya (Microsoft CEO) and others are very anxious. Murati: Very bad. Altman: Understood. Can you speed things up? Microsoft has been pressing me for a reply. Murati: Sam, it’s really very bad.

In just a few words, “very bad” was repeated twice. Murati’s tone was like giving a warning to a family member of a critically ill patient—not a cold or flu, but ICU level.

Altman began trying to intervene personally:

Altman: Can I come over? Murati: They don’t want you to come.

They don’t want you to come. Five words, erecting a cold wall of separation. This “they” referred to colleagues who had just sat with him on the board a few hours earlier—Ilya Sutskever, Adam D’Angelo, Tasha McCauley, Helen Toner.

Altman clearly realized the situation was serious. He lowered his stance, even suggesting resignation:

Altman: What do you think can improve the situation? If it helps, I can step down at any time. If they’re planning to launch aggressive lawsuits against me, I don’t know what to do. Can you tell them I just want to resolve this and talk to them?

“Anytime I can step down.” A founder just kicked out, now begging to leave more thoroughly. This isn’t humility; it’s a survival instinct amid chaos—he doesn’t understand what the board really wants.

And Murati’s reply ended Act One with a thunderclap:

Murati: They’ve already made a decision. Altman: Decided to let me go? Or is this a new development? Murati: Yes, they’ve decided to let you go.

“They’ve already made a decision.” No bargaining, no room for negotiation. It’s a death sentence read aloud.

But the subsequent dialogue revealed a split within the board’s psyche.

Altman asked in confusion:

Altman: Then why did they spend the whole weekend talking about wanting me back?

Throughout that weekend, the board announced his firing publicly, yet privately signaled—“Maybe you can come back.” This back-and-forth left everyone baffled. Murati’s response was even more deadly:

Murati: They’re appointing a new CEO.

So that’s it. Not “bring you back,” but looking for someone to replace you while keeping you as a backup.

Altman instantly understood:

Altman: Have they decided who it is? Can I tell Satya? Is it confirmed? Or are you just bringing Satya into the loop?

He began to realize he was no longer the negotiator but an obstacle—the board’s only goal was to move him aside; who would sit in that chair was irrelevant.

Then Murati uttered the darkest humor of the entire coup night:

Murati: The new person is that unknown Twitch guy.

“That unknown Twitch guy.” This was the CTO of OpenAI, describing the upcoming CEO who would replace Sam Altman on the night of the coup.

Altman immediately knew who she meant:

Altman: Emmett? (Emmett Shear, co-founder of Twitch) Murati: Yes.

Before this, Emmett Shear had almost no interaction with OpenAI. He wasn’t a top AI scientist, a seasoned executive, or even a board member. He was just a former gaming live-streaming platform founder. The board spent less than 24 hours scouring the vast Silicon Valley talent pool to find someone to save OpenAI.

Murati’s remark about “that rando Twitch guy” wasn’t sarcasm; it was a public execution of the board’s decision quality.

Even more absurdly, Emmett Shear himself was replaced after just a few days. The execution quality of this coup was evident.

The conversation deepened. Altman’s team began speculating about the true motives of the board—not performance, not capability, but something more hidden.

Altman: Are they planning to transfer the IP to Anthropic? The team suspected so.

Anthropic, founded by former OpenAI employees, is OpenAI’s most direct competitor. If the board really intended to transfer OpenAI’s intellectual property to a rival, it would be one of the most absurd “enemy-of-ourself” moves in business history. But this was just a guess; Altman had no evidence, only panic.

Yet Murati’s reply was colder:

Murati: They just don’t want AGI to be controlled by you (Just not your hand on agi).

“Just not your hand on agi.”
It’s not “you’re unqualified,” not “you made a mistake,” but—whoever you are, as long as you are you, it’s not allowed. This is a strategic personal veto. The board believed that AGI (Artificial General Intelligence), the most important technological achievement in human history, must not fall into Sam Altman’s hands.

This sentence is the core of the coup. All official statements about “lack of transparency,” “bypassing safety reviews,” are just surface explanations. The underlying logic is simple: who controls the power, controls the future. The board doesn’t trust Altman.

Altman’s response carried a sense of exhausted despair:

Altman: I don’t have to come back! They were just calling me back yesterday morning.

“I don’t have to come back”—this was his second time voluntarily giving up. But the board’s capriciousness made such surrender meaningless.

In the final part of the messages, Altman threw out a proposal that could change AI history:

Altman: Wait, I have an interesting idea. Murati: Still with Satya, go ahead. Altman: What if Microsoft acquires OpenAI? Would that align with the governance structure the board wants?

What if Microsoft directly acquires OpenAI?

This idea was only revealed today. On the night he was ousted from his own company, Altman’s thought wasn’t about how to return but about selling the entire company—selling to Microsoft.

There are two implications: first, Altman already believed that OpenAI’s current non-profit governance was beyond saving, and it was better to tear it down and rebuild; second, he was willing to “sell out” so that Microsoft could become the actual controlling party, bypassing the board’s restrictions.

But Murati’s reply was intriguing:

Murati: Satya is very tactful.

“Very tactful” is a polite way of saying it. Translated, it means: Microsoft CEO Nadella didn’t bite. Why?

Because at that time, Microsoft was already OpenAI’s largest shareholder (49% stake), but the non-profit structure prevented Microsoft from directly controlling the board. Acquisition could solve the control issue but would bring bigger problems—antitrust scrutiny, ethical controversies over the non-profit mission, and the looming lawsuit from Musk. Microsoft chose to maintain the status quo rather than take over amid chaos.

The conversation ultimately ended with Murati confirming that the board was aware she had “re-hired Altman,” but it was too late. That weekend, the board officially appointed the “unknown Twitch guy,” while Altman’s stay at Microsoft marked a 13-day exile.

These text messages, only revealed after two and a half years, are now becoming key evidence in Musk’s lawsuit against OpenAI.

Musk’s core claim is that OpenAI betrayed its non-profit mission, turning into a commercial tool for Microsoft and Altman. And the texts where Altman proposed “Microsoft acquiring OpenAI” directly support Musk’s argument—inside, someone was already considering selling out.

Moreover, Murati’s statement that “they just don’t want AGI to be controlled by you” exposes the personal targeting behind the board’s decision. Musk’s lawyers will ask: if the board thought Altman was unfit to control AGI, why did he still hold power after Microsoft’s acquisition? Isn’t that proof that OpenAI had completely capitulated to capital?

This lawsuit sees Musk claiming $180 billion, demanding a complete overhaul of OpenAI’s profit model. And these messages, like bullets, are being chambered one by one in court.

In fact, that chaotic weekend in 2023 left no winners. Altman was ousted for two days, then forced back by staff, but the scars of governance remain at OpenAI; all board members were ousted, Ilya Sutskever quietly left; Microsoft didn’t acquire the company then, but from that point on, their control over OpenAI became more covert; and users and the public only saw a billion-dollar company, like a high school club, deciding the future of AGI with a text message and a “rando Twitch guy.”

Silicon Valley’s most dramatic power struggle left no heroes, only a trail of chaos. These messages are the clearest fingerprints on that chaos.