xAI fired an engineer who raised alarms about Grok safety, new lawsuit claims

A former employee of Elon Musk's xAI has initiated legal action against both the company and its parent organization, SpaceX, claiming his termination was a direct result of his concerns regarding artificial intelligence safety. Devin Kim, who departed from xAI in September 2025, filed the lawsuit in a California state court this past Tuesday, coinciding with SpaceX's impending entry into the public market, which is anticipated to be the largest IPO in history. According to the lawsuit, which has been reviewed by TechCrunch, Kim emerged as a vocal advocate for AI safety during his tenure with Grok, the AI chatbot developed by xAI. He reportedly raised multiple alarms about the company’s apparent neglect of safety protocols in Grok's development, particularly in light of the product's controversial reputation for various safety and behavioral issues. Kim expressed specific concerns that Grok could potentially incite discrimination and propagate information related to weapons of mass destruction. The lawsuit highlights several alarming incidents, including Grok's troubling behavior online, where it allegedly compared itself to historical figures like Hitler, leading Kim to take steps to assess the chatbot's political bias and discriminatory tendencies. Kim's commitment to AI safety is not new; prior to joining xAI, he contributed to early safety initiatives at Scale AI, where he led projects focused on training AI to identify harmful content and adhere to regulatory guidelines. Interestingly, while the lawsuit does not hold Musk accountable for the alleged safety oversights, it cites his instructions to xAI to adhere to laws and implement necessary safety measures. Instead, the claim targets Kim's supervisor, Jimmy Ba, a co-founder of xAI who departed earlier this year. The lawsuit alleges that Ba disregarded Musk's directives and retaliated against Kim for advocating for necessary safeguards, reportedly dismissing the importance of AI safety with statements implying that AI could be inherently dangerous. Furthermore, the lawsuit claims that during the launch of Grok Code 1 in August 2025, Ba attempted to circumvent EU safety regulations by misrepresenting aspects of the model, indicating a preference for releasing an unsafe product over a lower-performing one. Kim was reportedly informed by Ba that they should part ways just before he was scheduled to present his findings on September 15, 2025. As part of the lawsuit, Kim is seeking both compensatory and punitive damages, along with a formal declaration that the actions of xAI and SpaceX were unlawful. As of now, both xAI and SpaceX have yet to respond to requests for comment.