SAN FRANCISCO — Elon Musk has now lost to OpenAI twice in a month. The second defeat, handed down on Monday, threw out his AI company’s claim that the rival stole the secrets behind its Grok chatbot, and the judge did it in a way that leaves xAI no room to try the same argument again.
US District Judge Rita Lin, in San Francisco, dismissed xAI’s trade-secret suit with prejudice, the legal term for a door closed rather than left ajar. The whole case had narrowed to one person: Xuechen Li, a former senior engineer at xAI whom OpenAI later tried to hire, the South China Morning Post reported. xAI argued that in courting Li, OpenAI was really fishing for the confidential work behind Grok.
The theory was specific. xAI’s amended complaint centred on a presentation Li gave while OpenAI was recruiting him, and claimed the company wanted the secrets tied to the July 2025 release of Grok 4 because it believed its own forthcoming ChatGPT update could not compete on complex reasoning. OpenAI, the suit said, was lagging in the reinforcement learning and post-training techniques Li understood, and saw an interview as a way to close the gap.
Lin did not buy it. She found xAI had failed to show that OpenAI induced Li to hand over anything confidential, or that its engineers even knew he might. Asking a job candidate to talk through prior work, she reasoned, is what hiring looks like everywhere, and a routine interview question is not evidence of a plot to steal. Having already tossed an earlier version of the complaint in February, she concluded a third attempt would be futile and ended the case for good, the suit, as Engadget put it, thrown out.
It is the second time in four weeks a court has told Musk no. On May 18 a federal jury rejected his far larger $150 billion claim that OpenAI and Sam Altman had betrayed the company’s founding nonprofit mission. The man who co-founded OpenAI and then spent years attacking it from the outside keeps arriving at the same result, and it is not the one his filings predict.
Strip away the billionaire feud and the ruling is really about something every AI lab is living through: the talent war. The frontier companies are a small world, and engineers move between them carrying knowledge that does not stay neatly behind a former employer’s firewall. Trade-secret law was not built for an industry where the most valuable assets walk out the door in someone’s head. Lin’s decision draws the line where most employment lawyers expected, which is that hiring a rival’s people, and asking them what they know, is competition rather than theft.
That distinction matters well beyond xAI and OpenAI, because if a recruiting conversation could be recast as misappropriation, every lab poaching from every other lab would be one disgruntled filing away from court. The danger for the loser is that the precedent cuts against you later. xAI does its own aggressive hiring, and a ruling that protects OpenAI’s recruiting protects everyone else’s too.
Musk, meanwhile, is fighting on more than one front. xAI is also a defendant in a separate dispute far from Silicon Valley, where thousands of Mississippi residents are suing over the gas turbines powering its data centres near Memphis. The legal calendar is filling up faster than the wins are.
What the dismissal does not resolve is the rivalry itself, which is being decided in the labs rather than the courtrooms no matter how the filings land. xAI cannot bring this claim again, and OpenAI walks away validated, but neither outcome tells you whose model reasons better next year. As for Xuechen Li, the engineer whose job interview became the centre of a lawsuit between two of the most valuable companies on earth, the ruling returns him to where most engineers prefer to be, which is a footnote.
