A legal battle involving Elon Musk looks shaky, as he charges OpenAI with unfairly snatching eight workers from xAI to sneak a peek at private details about its data setups and the AI chat system called Grok.
Judge Rita F. Lin from a US district court decided on Tuesday to approve OpenAI’s request to drop the case, pointing out that xAI didn’t offer any solid signs of wrongdoing by OpenAI.
xAI’s focus stayed mostly on what ex-workers supposedly did wrong, but when looking at the accusations, the judge noted that xAI couldn’t prove OpenAI pushed any of these people to take secrets or that those folks actually applied any taken info after joining OpenAI.
A pair of workers confessed to grabbing private data, including pulling down xAI’s core programming and one of them wrongly taking a private audio from a company-wide talk by Musk.
However, others faced claims of holding onto minor stuff like old work messages on their gadgets, or they appeared to have no secret info at all.
The judge highlighted especially thin points in xAI’s filing, which even admitted one hired person never got to see the private details after leaving, and two others were just tossed in because they switched jobs from xAI to OpenAI, as the judge mentioned.
Based on the slim details available, the judge wrapped up by saying that although xAI might have grounds to go after a few past workers for misuse, it doesn’t have a strong case against OpenAI for taking secrets.
The judge’s decision probably won’t wrap up the whole dispute, since she’s giving xAI a chance to fix up its filing and sort out the weak spots.
We weren’t able to get a quick response from xAI on this, so it’s not clear what their next move might be.
Still, it seems xAI won’t back down easily from this conflict, which OpenAI has called part of a bigger push by Musk to bother them through various court cases questioning how his top rival runs its operations.
As expected, OpenAI cheered the ruling on their X account, claiming this groundless legal action was just one more piece in Elon Musk’s ongoing effort to pester them.
People in other tech firms who are pulling in skilled people for AI work will probably feel better after seeing the judge’s words. A lawyer who handles business disputes, Sarah Tishler, explained that the decision comes down to a basic idea in laws about protected business info: bringing on staff from a rival doesn’t equal swiping their hidden knowledge.
To win under the law that protects trade secrets, xAI needs to demonstrate that OpenAI got hold of and applied those supposed secrets, not merely that they brought on board people who might have grabbed them, as Tishler put it. Things like odd timing, strong efforts to hire, or even files that got downloaded aren’t sufficient by themselves.
Tishler thinks this outcome will be good news for AI companies looking to grab top experts without worrying about court troubles from how they hire.
In the world of AI, where skilled people shift quickly and the rivalry is huge, this decision makes it clear that just having doubts isn’t adequate, Tishler added. You’ve got to prove the taken details really ended up with the other side and got used.
In this court fight, Musk argues that OpenAI breaks California’s rules on fair business by trying to wipe out real rivalry in AI through blocking xAI’s new ideas and making xAI battle against its own hidden info.
Yet, this argument depends completely on showing that OpenAI took those workers specifically to grab secrets. For the case to move forward, xAI has to strengthen the proof for its other accusation that OpenAI broke the national law protecting trade secrets, according to the judge. To pull that off, xAI must show OpenAI wrongly got, shared, or applied a secret without xAI’s okay.
Proving that could be tough, since right now xAI hasn’t given any real claims that OpenAI itself grabbed, spread, or used xAI’s hidden info, as the judge stated.
The only thing xAI has said is that OpenAI got ex-workers to pass along secrets, but so far there’s zero support for that, the judge explained. Tishler pointed out that the court also turned down xAI’s idea that OpenAI should answer for what new hires did prior to starting, for the same cause: without proof that OpenAI told them to steal or truly used the taken stuff, the company can’t be blamed.
The best proof xAI offered about a worker’s bad actions, which supposedly let OpenAI take xAI’s secrets, centers on the exit of one early xAI tech expert, Xuechen Li.
But even that didn’t cut it, the judge said. xAI claimed Li did a talk for OpenAI that included private details. Also, Li put the full xAI programming set into his own online storage, linked to ChatGPT, the judge mentioned, right after a hiring person from OpenAI messaged him on Signal with a link to some other unrelated online spot.
xAI wanted the court to be surprised by those Signal chats, hoping it would see hidden meanings like xAI did. To back up that OpenAI got the programming, xAI highlighted a Signal note from the OpenAI hirer sent four hours after Li took the code, which said “nw!”
xAI says this stands for “no way!”—meaning the hirer was excited about getting xAI’s code. But in a side note, the judge said OpenAI claims “nw” means “no worries,” so it’s not tied to Li choosing to upload the code to a spot connected to ChatGPT.
Even if going with xAI’s take on the message, xAI still didn’t prove the hirer or OpenAI looked at or asked for the files, the judge added.
xAI’s position got weaker because a short-term court order from another case against the tech expert stopped Li from taking the OpenAI job.
That order made OpenAI pull back its offer to Li. And that’s an issue for xAI, since Li never started at OpenAI, so obviously he didn’t apply xAI’s secrets there.
On top of that, if Li really passed private info in his job talk for OpenAI, xAI didn’t claim any details showing OpenAI knew Li was giving away xAI’s secrets, the judge wrote.
This setup makes it really difficult to say OpenAI ever applied what he supposedly took, Tishler shared with us.
One more ex-xAI tech person, Jimmy Fraiture, got blamed for duplicating xAI secrets, but Fraiture claims he erased the wrongly taken info before beginning at OpenAI. Crucially, the judge said, since he got on board with OpenAI, there’s no sign he used xAI secrets to help the competitor.
Apart from the simple fact that Fraiture got approached by the same OpenAI staffer who went after Li, xAI doesn’t claim any details showing OpenAI pushed Fraiture to grab xAI’s private info to start with, the judge wrote.
Because none of the other past workers supposedly passed to or told OpenAI about any xAI secrets, xAI couldn’t push forward its accusation that OpenAI took secrets just based on stuff linked to Li and Fraiture’s claimed bad moves, the judge explained.
xAI might fix its filing to keep these points going, but up to now the firm has shown only thin, fully indirect proof.
xAI could possibly find more support for its claims of misuse against OpenAI in its current court case with Li. We couldn’t get a fast reply from Li’s attorney to see if this new ruling affects that situation.
One of the weakest points xAI brought up was saying an unnamed money manager quit xAI for a lower spot at OpenAI after picking up all his knowledge on data setups from xAI.
That manager brushed off xAI when the firm tried later to ask about worries over private info.
The ex-xAI manager supposedly replied with “Suck my dick,” turning down any chance to say how his new role at OpenAI might connect to what he did at xAI. “Leave me the fuck alone.”
xAI attempted to say the manager’s anger proved wrong actions. But the judge wrote that xAI only said the manager “just had xAI secrets on data centers” and didn’t say he ever applied secrets to aid OpenAI.
If xAI had proof that OpenAI’s approach to data centers quickly copied xAI’s after the manager switched, that could have boosted xAI’s side. But lots of things could make a past worker ignore contact from an old boss after leaving, the judge hinted.
His anger when xAI contacted him about private worries doesn’t back up a real guess of using secrets, the judge wrote. Being mad at a previous boss when quitting doesn’t, by itself, show using secrets in the next role. Also, the manager’s zero background with AI data setups before xAI, without extra details, doesn’t back a solid guess that he used xAI’s secrets at OpenAI.
xAI has up to March 17 to update its filing and continue this specific battle with OpenAI. But the firm can’t toss in fresh accusations or people, the judge said, or switch up the claims except to fix the pointed-out problems.
For Li, the tech expert blamed for sharing xAI secrets with OpenAI, this court fight might cut down one area of questions as he deals with two more legal scraps over xAI’s secret claims.
Tishler has kept a close eye on xAI’s court fights over protected info. Back in October, she pointed out that Li is stuck in a tough spot, dealing with demands in the regular court from Musk to give up data that might hurt him in the FBI’s crime probe into Musk’s charges. As Tishler put it:
The real situation is tough: Li has to pick between staying quiet to shield himself in the crime case, and facing the fallout in the regular court from that choice. Say no to answering, and xAI might push for bad assumptions; answer, and those words could help the crime side.
In the end, the FBI wants to show Li took info that counts as a protected secret and planned to use it to help OpenAI, while aware it would damage xAI. If they do that, xAI would have official backing that its secrets got taken, Tishler wrote.
If xAI got that kind of boost and managed to revive the OpenAI case, the main issue in the lawsuit tossed out today would change, Tishler thought, from asking if a theft happened to figuring out what OpenAI knew and at what point they knew it.
Other Stories You May Like