Judge skeptical of xAI claims that OpenAI stole trade secrets

SAN FRANCISCO (CN) — Elon Musk’s artificial intelligence company told a federal judge Tuesday that claims against OpenAI should not be dismissed because there is “plausible evidence” OpenAI was involved in stealing trade secrets.

Musk’s xAI sued in September 2025 over claims that OpenAI hired former xAI employees with knowledge of confidential information in a “strategic campaign” against the company. They add that OpenAI “poached” at least eight xAI employees in its “deliberate scheme” against the company because it was threatened by Grok, which competes with OpenAI’s ChatGPT.

“This is several employees all at the same time, all stealing information, all talking to the same recruiter, and all pivoting to OpenAI,” xAI attorney LeElle Slifer of Winston & Strawn said. “If you look at all of these facts together, it gives the plausible inference that OpenAI was involved, knew about it, and directed it.”

Last week, U.S. District Judge Rita Lin issued a tentative ruling dismissing the case against OpenAI while allowing xAI leave to amend. At Tuesday’s hearing, the Barack Obama appointee asked xAI how she can draw a line between the actions of specific former employees and OpenAI coordinating the scheme.

“Help me understand how it is I should infer OpenAI is directing people to take confidential information,” the judge prompted lawyers for xAI.

Musk’s argument focused on two former xAI employees, Xuechen Li and Jimmy Fraiture, whom the tech billionaire accuses of bringing confidential information to OpenAI.

Attorneys for xAI claim that Li downloaded source code while being recruited by OpenAI and disclosed trade secrets during a presentation in the interview process. They also accuse Fraiture of airdropping himself source code onto a personal device and stealing a nonpublic video of Musk the day after he accepted a job offer from OpenAI.

Slifer argued that OpenAI’s secretive communications with Li over the encrypted messaging app Signal, plus the company using the same recruiter to target both Li and Fraiture, is “indicative” that OpenAI was trying to steal information. However, she acknowledged they don’t have all the information yet because xAI has not yet been allowed to move forward to discovery.

“One or two, maybe is a coincidence; this is a pattern," Slifer said. “What do all of these individuals have in common? They have no problem lying to xAI, stealing information. It is further evidence that OpenAI is targeting individuals willing to breach confidentiality.

“When you add that on top of the mountain on the other side, it’s one more piece of the puzzle,” she added. “Discovery will allow us to fill it in.”

In contrast, OpenAI argued that xAI need to have actual facts before they get discovery to “rummage around in competitors’ files.”

Carolyn Luedtke, an attorney for the defendants from LA-based Munger, Tolles & Olson, said xAI “raised a lot of concerns” about Li and Fraiture; however, there are no claims that link their actions to OpenAI.

“OpenAI was not on notice of any potential wrongdoing, nor is there any evidence OpenAI induced or encouraged or said ‘we would like you to bring us trade secrets.’ There is just speculation, and speculation is simply not enough,” Luedtke said.

Two of the former xAI employees did nothing wrong, she added, and two others “kept some documents” on their personal devices after leaving xAI. Another tried to use a personal device to access a blocked link.

As for Li and Fraiture, Luedtke said, Li was never hired by OpenAI, and Fraiture deleted any confidential information after he took a job at OpenAI.

“Those don’t show any coordinated effort,” Luedtke said. “The fact that lots of people left is just as much evidence that this place was a bad place to work. The weaker the strands, the weaker the case. Here, their strands are pretty weak.”

In rebuttal, Slifer accused the defendants of minimizing the allegations against the former employees. She explained that OpenAI hired Li, but he did not get the chance to work because of a previously ordered injunction. Additionally, she said that Fraiture has lied to xAI in the past, so his claims that he deleted any confidential information aren’t believable.

“This was not disgruntled employees; they were stealing the entire source code,” she said.

The judge further questioned xAI about the additional lawsuits they have filed against the individual former employees, saying it was “particularly odd” to ask for discovery to make claims against OpenAI in this case when they have the opportunity for discovery in other cases.

Slifer responded by saying xAI “should not be punished” for seeking additional litigation against the ex-employees in their individual capacities.

Lin did not indicate when she would release a ruling.

Representatives for both parties did not immediately respond to a request for comment.

Last August, xAI filed a lawsuit against Li, claiming the engineer stole confidential information and trade secrets. The next month, Lin ordered Li to turn over his personal electronic devices and temporarily blocked him from working at OpenAI or another competitor of xAI concerning generative AI, as well as communicating about generative AI.

Musk, an early investor and cofounder of OpenAI, also sued OpenAI CEO Sam Altman and his company in 2024, seeking a court order to prevent it from going through with for-profit restructuring. Musk claimed the move violated OpenAI’s nonprofit mission and breached the terms of his previous donations to the company, approximately $45 million. A jury trial in the case is tentatively scheduled for March 30.

Musk and Altman, who cofounded OpenAI in 2015 and later clashed over who should lead it, have had a long-running feud over the startup’s direction ever since Musk stepped down from the company board in 2018.

Follow @mhattridge_