Musk’s “World War III” threat in Twitter lawsuit haunts him at OpenAI trial

Just days before the trial started, Elon Musk tried to settle his lawsuit, which alleges that under Sam Altman’s direction, OpenAI abandoned its mission to serve as a nonprofit making AI to benefit humanity.

According to a Sunday court filing from OpenAI, Musk messaged OpenAI President Greg Brockman two days ahead of the trial to “gauge interest” in a possible settlement. Brockman promptly responded, suggesting that “both sides” drop their claims. But Musk refused, then appeared to grow threatening enough that the court may allow Brockman to testify on the message as evidence supposedly revealing Musk’s true motives for pursuing the litigation.

“By the end of this week, you and Sam will be the most hated men in America,” Musk responded to Brockman’s suggestion that all claims be dropped. “If you insist, so it will be.”

OpenAI clearly did not accept the settlement terms, as the trial started last week with Musk as the first witness. On the stand, Musk stumbled several times, perhaps weakening his case by making concessions, growing hot-tempered, backing off claims that AI risks may quickly become existential, and admitting his ignorance when it comes to AI safety at his own company, xAI.

If admitted, his alleged threat could become his next big stumble, as Brockman—whom Musk also wants out at OpenAI—will be allowed to testify about the message when he takes the stand, likely today and tomorrow.

Typically, communications during proposed settlements aren’t admissible, but likely extra-frustrating to Musk, OpenAI pointed to an exception made during Musk’s failed lawsuit attempting to back out of his Twitter purchase.

In that 2022 case, Musk’s legal team invited a “renegotiation” of the Twitter purchase price “so that the lawsuit could be dropped,” while threatening that “it would be World War III until the end of time for real” for Twitter leaders and “their heirs,” if Musk was forced to buy Twitter at a price Musk set based on a 420 joke. During that exchange, Musk also supposedly tried to make Twitter executives uncomfortable by reminding them that if he “ends up owning this thing, he’ll have access to all of the company’s records and he could look at everyone’s emails and dig into whatever he wanted to dig into.”

That exchange ended up being admissible because Musk’s legal team intended to disclose the threat to opposing counsel. OpenAI suggested that Musk’s message to Brockman should similarly not be viewed as privileged or a sincere effort to settle the litigation because it’s “coercive rather than conciliatory.”

OpenAI alleged that the threat was in line with “similar menacing statements” Musk has made throughout the litigation, which they claim is little more than a harassment campaign attempting to eject Altman due to Musk’s personal grudge against the OpenAI CEO. As OpenAI has alleged, Musk’s message should be admitted because it “tends to prove motive and bias,” specifically proving that “Musk’s motivation in pursuing this lawsuit is to attack a competitor and its principals.”

For Musk, the attempt to get his message to Brockman admitted is probably yet another unwelcome flashback to the Twitter case he lost. However, it seems unlikely that it wouldn’t come up, since OpenAI’s lawyer who grilled Musk on the stand, William Savitt, was on Musk’s legal team when he tried to back out of buying Twitter. Savitt’s memory of that “World War III” threat was likely fresh at the moment when the message to Brockman was delivered.

Whether Judge Yvonne Gonzalez Rogers will admit the email has yet to be determined, but she will probably have to decide today, ahead of Brockman’s testimony starting. Musk’s lawyers are likely hoping the judge rejects OpenAI’s arguments and strictly adheres to precedent requiring that communications during settlement negotiations aren’t admissible, since allowing it could cloud sincere settlement negotiations in the future.