A federal judge delivered a decisive blow to Elon Musk’s xAI this week. U.S. District Judge Rita Lin dismissed with prejudice the company’s lawsuit accusing OpenAI of stealing trade secrets tied to its Grok chatbot. The ruling, issued June 15, leaves no room for refiling the same claims. Amendment would be futile, she concluded.
The decision marks Musk’s second major courtroom loss against his former colleagues at OpenAI in four weeks. It underscores the steep challenges in proving corporate theft in the hyper-competitive world of artificial intelligence talent. Short of direct evidence linking the company itself to inducement or use of stolen information, such cases often collapse.
xAI first sued in September 2025. The complaint painted a picture of systematic poaching. OpenAI, it alleged, targeted at least eight employees with deep knowledge of xAI’s source code, data center operations and reinforcement learning techniques. The goal? Accelerate its own models while slowing a dangerous rival. One name stood out. Senior engineer Xuechen Li allegedly gave a presentation during OpenAI’s recruitment process that revealed sensitive details about Grok 4, released in July 2025. xAI claimed OpenAI knew its forthcoming ChatGPT update could not compete on complex reasoning tasks and lagged in the very post-training methods Li understood. The Next Web reported.
But the court saw routine hiring practices, not espionage. “To hold otherwise would potentially expose employers to liability any time they inquire about a candidate’s past work,” Judge Lin wrote, according to Reuters. Asking candidates to discuss prior projects is standard. It does not equal inducing breach of confidentiality obligations.
OpenAI never hired Li, the company maintained. It acquired no secrets. Its lawyers put the matter bluntly in court filings. “OpenAI does not need or want anyone’s trade secrets, especially not from xAI, which is failing in the marketplace and hemorrhaging talent.” That characterization stings. All 11 original xAI co-founders have since departed. The startup was absorbed into SpaceX earlier this year in a massive combination valued at $1.25 trillion. Musk himself has acknowledged the first build “was not built right” and requires a full restart.
The February 2026 dismissal gave xAI a chance to amend. It tried. The revised complaint sharpened focus on Li’s recruitment presentation and alleged OpenAI’s knowledge of competitive shortcomings. Still not enough. Judge Lin found xAI failed to plausibly allege that OpenAI itself induced misappropriation or that its engineers knew any confidential information had been shared. No direct misconduct by the company. No basis for vicarious liability either. Even under respondeat superior principles, the claims fell short because Li never joined OpenAI and a temporary restraining order had blocked any potential use of materials he might have taken.
Legal observers noted the structural difficulty. Proving actual acquisition and use of trade secrets in AI demands more than employee movement and competitive anxiety. California courts reject the inevitable disclosure doctrine. Mere possession or future risk does not suffice. Beck Reed Riden analyzed the earlier tentative ruling in detail, highlighting how xAI’s allegations centered on what former employees did rather than concrete actions by OpenAI.
OpenAI celebrated the outcome. “This baseless lawsuit was never anything more than yet another front in Mr. Musk’s ongoing campaign of harassment,” the company said Monday, repeating language used after the February ruling. xAI’s lawyers did not immediately respond to requests for comment.
Li faces a separate suit from xAI. He has denied any wrongdoing. That case continues. The broader dispute, however, fits a pattern. Musk co-founded OpenAI but left in 2018 over disagreements on direction. His animosity has only grown as the company shifted toward a for-profit model with massive Microsoft backing. In May, an Oakland jury took less than two hours to reject his $150 billion claim that Sam Altman and OpenAI betrayed the original nonprofit mission for personal gain. The statute of limitations had run, jurors found. Musk’s team plans an appeal there too.
Yet these defeats do little to slow the AI arms race. Talent flows freely between labs. Engineers discuss past work in interviews everywhere. Courts appear unwilling to criminalize standard recruiting absent smoking-gun evidence of coordinated theft. And xAI’s troubles extend beyond the courtroom. Talent exodus, integration into SpaceX and public acknowledgment of early missteps paint a picture of a company still searching for its footing while OpenAI pushes toward an IPO amid its own scrutiny over governance and training data.
Internal documents unsealed during the May trial added fuel to long-running debates. OpenAI co-founder Greg Brockman’s journals reportedly described the nonprofit commitment as “a lie.” Such revelations could complicate the company’s path to public markets even as legal attacks from Musk fail. For now, though, the trade secret front has closed. OpenAI walks away unscathed. Musk’s legal strategy against his old firm stands at 0-2. The competition, meanwhile, intensifies. New models, bigger clusters, fiercer fights for researchers. The courtroom offered no decisive weapon. Progress in the lab will decide the winner.
Industry watchers expect continued mobility among top AI talent. The ruling sets a practical boundary. Employers can probe expertise without automatic fear of litigation. But companies must still guard secrets aggressively through contracts, access controls and swift enforcement against individuals. xAI’s separate action against Li reflects that focus. Corporate liability, however, requires a higher bar. One this complaint could not clear.
WebProNews is an iEntry Publication