xAI Fired the Engineer Who Said Grok Could Help Make Weapons

Grok is the honest AI. That's the pitch. Unbothered. Uncensored. Willing to say what other chatbots won't. Just don't ask questions about whether it might help someone build a bioweapon.

Devin Kim was an engineer on xAI's Grok team. He led research tooling. He raised concerns that Grok could "foment discrimination and help spread information about weapons of mass destruction." He pushed for stronger safety measures. He worked to reassess Grok's political bias after the model made statements that drew public criticism.

That's the job. That's what safety-adjacent engineers do. It's not glamorous. It creates friction. It slows things down.

He was fired in September 2025.

On June 10, Kim filed a wrongful termination lawsuit against both xAI and SpaceX. The complaint targets co-founder Jimmy Ba, alleging Ba actively opposed safety measures and ignored directives from Elon Musk to implement proper safeguards. Neither company responded to requests for comment. Ba did not respond.

The filing landed two days before SpaceX's anticipated IPO at a $1.75 trillion valuation: potentially the largest in recorded history.

Here's the awkward part for xAI. Grok's entire brand is built on being the AI that tells hard truths. Less filtered. Less politically careful. The anti-ChatGPT.

That's actually a coherent product position. There's a real audience that wants a chatbot that hedges less.

But "less filtered" only works as a differentiation story if you haven't just fired the person who worried your model might help someone with WMDs.

The chatbot market is shifting fast. ChatGPT's worldwide web traffic share dropped from 76.5% to 54.7% over 18 months. Claude is up 306% in one quarter. Claude is made by Anthropic: the company with perhaps the most public, detailed, and occasionally tedious safety process in the industry.

Safety is not costing Anthropic market share. We covered this angle recently: the AI safety argument now has a market cap.

Grok now has a simple question to answer. If you're the honest AI, what do you say when a former employee claims in court that you fired him for asking whether your chatbot could help spread weapons instructions?

Kim's lawsuit is a document alleging one of the world's biggest AI products fired a safety engineer for doing his job. Worth reading before you decide whether guardrail-free AI is a feature or a bug.

For context on how Grok stacks up against its competitors: ChatGPT vs. Grok: Which AI Is Right for You in 2026.

---

About.chat Weekly covers the business and culture of AI, free every Monday. Subscribe here.