The frantic late-night text messages that Sam Altman fired off as he scrambled to claw back his job at OpenAI have been laid bare in a Californian courtroom, offering an extraordinary glimpse into the 48 hours that nearly toppled one of the world’s most valuable technology companies.
The exchanges between Mr Altman and Mira Murati — the company’s chief technology officer who had abruptly been installed as interim chief executive — were shown to jurors at the Musk v Altman civil trial taking place in a federal courthouse in Oakland. They cover two days in November 2023, when the OpenAI board’s shock decision to fire its co-founder briefly threatened to splinter the firm behind ChatGPT.
According to court documents seen by Business Insider, the conversation begins at 5.43pm Pacific Time on 19 November 2023, with Mr Altman apparently locked out of OpenAI’s offices and asking Ms Murati to “officially invite me to the office for a meeting.” She tells him she is “about to speak with them” — a reference to the board members who had just removed him.
What follows reads less like the correspondence of a Silicon Valley titan than the desperate messaging of a man watching his life’s work disappear in real time. Around seven hours later, with the clock past 2am on 20 November, Mr Altman pings her again for an update, asking whether she can indicate “directionally good or bad”. He notes that “Satya” — Microsoft chief executive Satya Nadella, whose company had ploughed billions into OpenAI — was “concerned”.
The replies coming back from Ms Murati are unsparing. “They’re convinced about their decision,” she writes. When Mr Altman asks whether that decision is “for me to be fired?” she confirms: “Yes, for you to be gone.” She then breaks the news that she, too, is being replaced. “They want a new ceo in place,” she tells him.
Pressed for details, Ms Murati reveals the identity of her successor with one of the more memorable shorthand descriptions to surface in modern tech litigation: “is rando twitch guy”. The reference is to Emmett Shear, the former Twitch co-founder who would briefly hold the top job at OpenAI before Microsoft’s intervention helped engineer Mr Altman’s return. Mr Shear, evidently amused, changed his banner on X this week to read “new guy is rando Twitch guy” using screenshots of the disclosed messages.
The pleading from Mr Altman is relentless. “Then can i come in and talk about a path forward with them?” he writes. The board, Ms Murati replies, are saying no and need more time. Just ten minutes later he tries again: “still don’t want me?”
He asks at one point whether the board may be plotting to transfer OpenAI’s intellectual property to a rival. “Is that what they want is the IP going to anthropic?” he writes — a reference to the AI safety lab founded by former OpenAI staff. “That’s what the team thinks,” Ms Murati responds, before adding that the directors had become firm on one specific point: that Mr Altman should have nothing to do with the company’s pursuit of artificial general intelligence. “Just not your hand on agi,” she writes.
“I can not come back!” Mr Altman replies.
As the texts continue, he probes the possibility of Microsoft swallowing the company entirely. “what if msft acquires openai?” he asks at one stage. Ms Murati at another point expresses hope that Mr Nadella might “help undo this”, though she warns that even a mass walkout might not move the board. “They don’t care if everyone quits,” she writes.
The exchange ends with Mr Altman floating the strategy that, within hours, would in fact see him reinstated: a co-ordinated petition. “I think you all just need to get a petition of everyone saying they will quit and join” Microsoft, he tells her. “I’m on board,” she replies.
Later that day, around 600 OpenAI employees — about 95 per cent of the company’s staff — signed an open letter threatening to follow Mr Altman to Microsoft unless the board resigned and reinstated him. The strategy worked. Mr Altman was back as chief executive within days, and the directors who had fired him departed.
The texts have re-emerged not because of fresh corporate drama but because they form part of the documentary evidence in Elon Musk’s civil suit against Mr Altman, OpenAI and its president Greg Brockman. Mr Musk, who co-founded OpenAI in 2015 and was once one of its most prominent backers, is suing over the company’s controversial conversion from a nonprofit research outfit into a hybrid for-profit structure. He alleges breach of charitable trust and unjust enrichment, and is seeking, among other things, the removal of Mr Altman and Mr Brockman from their roles.
The case is being heard before US District Judge Yvonne Gonzalez Rogers and a nine-person jury, although their role in the so-called liability phase will be advisory; the judge will issue the final ruling. The trial is now in its second week. According to coverage by CNBC and the BBC, OpenAI is currently valued at around $852 billion, while Mr Musk has separately built up a competing AI venture, xAI.
Tensions between the two men, once close allies, have spilled out repeatedly into public view. Mr Musk has taken to referring to Mr Altman online as “Scam Altman”, while Mr Altman has not been above goading him in return.
For all the courtroom theatre, however, it is the Murati-Altman texts that have captured the public’s attention this week. They land somewhere between a corporate thriller and a workplace tragicomedy: a 38-year-old man pressing send into the small hours, asking the same question of an ally on the inside, and getting the same brutal answer. They don’t want you.
