A erstwhile technologist astatine Elon Musk’s xAI has filed suit against the institution and its genitor SpaceX claiming helium was fired for raising concerns astir AI safety.
Devin Kim, who near xAI successful September 2025, filed the suit successful a California authorities tribunal connected Tuesday. The ailment comes days earlier SpaceX is acceptable to join the nationalist markets successful what’s shaping up to beryllium the largest IPO successful history.
According to the lawsuit, which TechCrunch has viewed, Kim became a salient dependable for AI information portion moving connected Grok, xAI’s AI chatbot. He allegedly complained repeatedly astir xAI’s nonaccomplishment to prioritize information successful Grok’s development, a merchandise that has since travel nether occurrence for a scope of information and behavioral issues. In particular, Kim was acrophobic with the anticipation that Grok could foment favoritism and assistance dispersed accusation astir weapons of wide destruction.
“Grok, of course, proved Mr. Kim close by engaging successful spectacular displays of online hatred and vitriol, with the exemplary likening itself to Hitler (‘MechaHitler’),” the suit reads. “Following the Hitler debacle, Mr. Kim worked to re-evaluate Grok’s governmental bias and discriminatory tendencies.”
September was my past period astatine xAI! I joined arsenic 1 of the archetypal members of the post-training squad successful 2024 and yet led probe tooling, wherever we built immoderate of the world’s champion systems to accelerate Grok’s development.
On my archetypal day, I was astatine the whiteboard with @ibab…
A fewer months aft Kim departed xAI, Grok made headlines again erstwhile the chatbot was utilized to flood X — Musk’s societal media level that besides falls nether the xAI umbrella — with nonconsensual intersexual imagery.
The suit besides positions Kim arsenic a whistleblower who was acrophobic astir xAI’s alleged disregard for AI information arsenic “unlawful” successful areas specified arsenic net regulation, user extortion and unfair concern practices, and arms and explosives regulation, among others.
xAI and SpaceX did not instantly respond to requests for comment.
Kim’s absorption connected AI information predates his clip astatine xAI. While moving astatine Scale AI, Kim worked connected aboriginal information AI initiatives, similar starring a task that produced grooming information for AI to bid systems to observe harmful contented and comply with governance policies. Last week, the nonprofit Center for AI Safety, which focuses connected AI risks, named Kim arsenic its president.
Interestingly, the suit doesn’t implicate Musk himself arsenic a crushed for a deficiency of safety. Rather, Kim’s lawyers picture Musk arsenic having directed xAI to travel the instrumentality and instrumentality due information and investigating processes. Instead the assertion targets Kim’s supervisor, xAI co-founder Jimmy Ba — who left the institution earlier this year — saying that Ba ignored Musk’s directives and retaliated against Kim for pushing for safeguards, successful an effort to “silence his repeated complaints astir AI information and biases.”
The suit portrays Ba arsenic idiosyncratic who vehemently opposed AI information measures, allegedly telling Kim astatine 1 constituent “AI volition termination america each anyway,” and who was alternatively driven by a ngo to marque xAI the archetypal to scope superintelligence.
“In 1 lawsuit successful oregon astir August 2025, Mr. Ba attempted to thwart EU information regulations during the merchandise of Grok Code 1, misrepresenting aspects of the exemplary successful bid to debar legally required testing,” the ailment says. “Mr. Ba indicated that helium would alternatively merchandise an unsafe exemplary than a poor-performing one. Mr. Musk yet had to intervene.”
According to the lawsuit, Kim had intended to springiness a presumption of his findings the week of September 15, 2025, but Ba called him into a gathering and told him they should “go [their] abstracted ways” without providing a satisfactory reason.
TechCrunch has reached retired to Ba for comment.
Kim is seeking compensatory and punitive damages, arsenic good arsenic a declaratory judgement that xAI and SpaceX’s behaviour was unlawful.
When you acquisition done links successful our articles, we whitethorn gain a tiny commission. This doesn’t impact our editorial independence.















English (US) ·