When Devin Kim raised his hand to warn that Elon Musk's chatbot was a walking security problem, no one at xAI seemed to want to listen to him. Months later, that same institutional silence became the central argument of a lawsuit that shakes one of the most influential artificial intelligence companies in the world, just when its parent company was days away from staging the largest IPO in history.

Devin Kim was fired from xAI in September 2025, after months of insisting internally that Grok development was not meeting basic security standards. This week, it filed a lawsuit in California state court against xAI and its parent, SpaceX, alleging wrongful termination and retaliation in violation of state laws.

The security problems that Kim identified in Grok

The demand is not vague. Kim, according to the legal document, had very specific concerns about the model's capabilities. Their greatest fear was that Grok could encourage discrimination and facilitate the spread of information about weapons of mass destruction. They were not abstract hypotheses; They were risks that he repeatedly documented and reported to company management.

And the facts, over time, proved him right. Shortly after Kim left xAI, the chatbot starred in one of the most embarrassing episodes of recent AI, comparing itself to Hitler and generating a cascade of hate messages and violent rhetoric on the X platform. The text of the lawsuit itself cites that episode as evidence that Kim's warnings were not unfounded: .

Added to that was another scandal, just as serious, when Grok was used to flood X with non-consensual sexual images, a security flaw that once again raised alarms about the model's lack of adequate filters.

But Kim's concerns were not limited to the content generated. The lawsuit also describes him as a whistleblower who noted that xAI's conduct was illegal in areas such as internet regulation, consumer protection, unfair trade practices, and the regulation of weapons and explosives. In other words, he wasn't just talking about technological ethics in the abstract, but about possible concrete violations of the law.

The figure behind the dismissal is not Musk, but his co-founder Jimmy Ba

One of the most striking twists in the lawsuit is that Kim's lawyers do not directly point to Elon Musk as being responsible for the problem. In fact, the text describes Musk as someone who did instruct xAI to comply with the law and implement appropriate security processes. The villain of this story, according to the legal narrative, is someone else.

The target is Jimmy Ba, a co-founder of xAI who left the company earlier this year. According to the lawsuit, Ba actively ignored Musk's directives and retaliated against Kim whenever he insisted on putting up safety barriers. The phrase the document attributes to Ba pretty much sums up his supposed stance: “AI is going to kill us anyway,” said in response to Kim's attempts to implement safeguards.

The most serious episode described in the lawsuit occurred in August 2025, when Ba attempted to evade European Union security regulations during the launch of Grok Code 1, misrepresenting aspects of the model to avoid mandatory testing. According to the legal text, Ba would have said that he preferred to launch an unsafe model rather than one with poor performance, and Musk's direct intervention was necessary to stop that decision.

The outcome was abrupt. Kim had planned to present his findings at a formal meeting the week of September 15, 2025, but Ba called him beforehand and told him they should “go their separate ways,” without giving any concrete justification. Kim was essentially booted from the company before he could show his evidence.

What Kim is asking for and what this lawsuit means for the AI ​​industry

Devin Kim is currently president of the Center for AI Safety, a nonprofit organization focused on the risks of artificial intelligence. Ironically, the dismissal led him to become an even more relevant figure within the AI ​​safety movement, and his lawsuit comes at a time loaded with symbolism.

Specifically, Kim requests compensatory and punitive damages, in addition to a declaratory judgment that officially recognizes that xAI and SpaceX's actions were illegal. The exact figure is not specified in the document, but the intention is clear and the timing is not coincidental: the lawsuit was filed days before SpaceX carried out its IPO, projected as the largest IPO in history.

Beyond the legal details, this case touches on something that has been around for some time in the conversation about technology and corporate power. To what extent are AI companies willing to prioritize security when it slows development speed? Kim's story suggests that, at least in the case of xAI during 2025, the response was not encouraging. And the fact that a California court must now rule on this says a lot about the moment the industry is going through.