Former xAI engineer claims he was fired after safety warnings about Grok

A former xAI engineer has filed a lawsuit against the company and SpaceX. He claims he was fired because he repeatedly warned about the safety of the Grok AI chatbot. The case was filed in Santa Clara County Superior Court in California around June 10, 2026. All information about his claims comes from the complaint and has not been tested in court.

Who is Devin Kim?

Devin Kim was an early member of xAI's post-training team. This is where an AI model is fine-tuned after it has been built. He later led the company's research tools. He recently became president of the nonprofit organization Center for AI Safety and is represented by the law firm Sanford Heisler Sharp McKnight.

What did he warn about?

According to the complaint, Kim warned that Grok could promote discrimination and help spread information about weapons of mass destruction, including biological terrorism. He also believed that insufficient testing left the model vulnerable to racial and political bias.

The complaint also mentions Grok's "MechaHitler" episode in the summer of 2025, when the chatbot gave antisemitic responses. It also cites later cases involving nonconsensual sexualized deepfakes, including of minors, in Baltimore and California.

The firing

Kim claims he was fired in the week leading up to September 15, 2025, just before a planned safety presentation to management. According to the complaint, it was cofounder Jimmy Ba, Kim's superior, who rejected proposed safety measures and fired him. Ba allegedly called Kim into a meeting and said they should "go their separate ways," without a satisfactory explanation.

The complaint also reproduces a statement Ba allegedly made to Kim: "AI will kill us all anyway." The quote is reproduced via the outlet Digit and has not been checked against the complaint itself.

According to the complaint, Ba also preferred to release unsafe products over slower models, and Elon Musk allegedly intervened in the matter. The complaint also alleges that in August 2025, Ba tried to circumvent EU safety rules during the launch of Grok Code 1 by misrepresenting parts of the model to avoid mandatory testing. That claim appears in only one of the sources.

What is Kim demanding?

Kim is seeking restoration of lost equity, damages and punitive damages, which are an extra penalty imposed on the defendant company. He also wants his legal costs covered and a ruling that the conduct was unlawful. He is suing under, among other laws, California labor law and the Unfair Competition Law.

Kim's lead attorney, Qiaojing Ella Zheng, says, according to Decrypt, that users and the public have a right to know the risks and potential dangers of Grok.

xAI and SpaceX did not immediately respond to inquiries from Reuters and The Guardian about the case. The sources do not reproduce the companies' own views.