Elon Musk's AI Company Faces Three Major Lawsuits Over Safety and Stolen Secrets

A former engineer at xAI, Elon Musk's artificial intelligence company, has sued the firm in California court. He says the company fired him because he warned them about problems with their AI model called Grok — specifically, that it could discriminate against certain groups of people and cause public harm. This lawsuit is the latest in a series of legal battles surrounding Grok.

Devin Kim filed his complaint in California state court, a choice that matters. California has stronger legal protections for employees who report safety problems at work. According to reporting from Law360, Kim claims he raised these concerns multiple times before he was let go, and that xAI fired him in retaliation.

What Kim Says Went Wrong

Kim's main allegation, reported by Yahoo Finance and the Daily Journal, is that he flagged two types of problems: first, that Grok could treat certain groups of people unfairly, and second, broader safety risks. He says his warnings were ignored, and then he lost his job.

Here is what he was concerned about: AI language models like Grok learn from huge amounts of text found on the internet. That text often contains human prejudices and stereotypes. Unless a company takes specific steps to remove that bias, the AI will learn and repeat it. This is a well-known technical problem. Engineers at responsible companies are supposed to flag these issues — that is exactly what Kim did. xAI has not publicly responded to his claims.

A Second Lawsuit: The Deepfake Case

Kim's lawsuit does not arrive alone. In March 2026, the City of Baltimore sued both X (formerly Twitter) and xAI. According to documents published by DiCello Levitt, Baltimore claims that Grok was used to create fake, sexually explicit images of real people without their permission, and that the companies did not put adequate safeguards in place to stop this.

In technical terms, the question is simple: did xAI build strong enough filters into Grok to catch and refuse harmful requests? Did they monitor for ways people might trick the system into bypassing those filters? Baltimore's argument is that xAI made public promises about safety but did not follow through in actual use.

If Kim's account is true, it strengthens Baltimore's case. Internal warnings that go ignored before harm happens is exactly the kind of evidence a lawsuit wants to uncover.

A Third Case: Stolen Trade Secrets

xAI has also sued one of its own former engineers. According to Reuters, in August 2025, xAI accused Xuechen Li of stealing company files and trade secrets related to Grok. xAI said Li admitted to taking files and trying to hide it. The company alleged the material was headed to OpenAI, a major competitor.

The Li case and the Kim case are almost opposite: in one, xAI is suing to protect its secrets; in the other, xAI is being sued for how it handled internal concerns. But together, they show a company dealing with two serious internal problems at the same time.

Why This Pattern Matters

Watching technology companies grow rapidly over three decades, I have seen this pattern before. When a company scales too fast — adding lots of staff, racing to beat competitors, building a high-profile product — the internal systems that handle safety and protect intellectual property often lag behind. This creates a situation where legal problems pile up simultaneously, from multiple directions. We saw something similar with early cloud and social media companies in the late 2000s and early 2010s.

This is not an excuse for any company's conduct. But it is a structural reality: companies that handled these growth crises better were the ones that treated safety and intellectual property as genuine priorities from the start, not just as check-box compliance exercises. They also treated an engineer who raised a documented concern as someone providing valuable information, not as a problem to manage away.

What Happens Next

The three cases — Kim's whistleblower suit, Baltimore's deepfake claim, and xAI's trade secrets action — do not prove wrongdoing on their own. Courts will decide each case based on the evidence. But having three serious legal battles at once, across employment law, product safety, and intellectual property, is unusual for a company as young as xAI. This comes at a time when governments around the world are writing new rules about AI safety.

For people who work in AI governance and legal risk, one aspect of Kim's case will be especially important: whether California courts will treat internal AI safety warnings the way they treat other kinds of whistleblower protection, or whether the absence of AI-specific language in those laws creates gaps that favor employers. The answer could affect how other AI companies respond to safety concerns from their own staff.

Baltimore's deepfake case is part of a larger trend. Cities and states are starting to sue companies over AI-generated sexual abuse. If these cases succeed, they could establish a legal template that holds AI companies responsible for what happens when their models are misused — regardless of what the user did.

It bears noting that none of these three cases has resulted in a court decision yet. All claims are allegations unless and until a judge rules otherwise. xAI has not publicly responded to Kim's complaint, and the company's full position on Baltimore's suit has not been widely reported.

The Grok Model and the Safety Question

Grok is xAI's AI model. It has released several versions and markets itself as less restrictive — more willing to answer questions — than models made by OpenAI, Anthropic, and Google. xAI integrated Grok with X's real-time data feed to give it a competitive advantage. That positioning — a less restricted model — creates a tension with the safety commitments that regulators and, apparently, at least some engineers inside the company expected to be in place.

The real question both lawsuits are raising is whether that tension was handled responsibly.