Days before SpaceX’s $75 billion initial public offering, a new lawsuit reveals that the engineer who raised the alarm on xAI‘s flagship chatbot Grok was fired to silence him — and that his supervisor dismissed every safety concern with a single grim maxim: “AI will kill us all anyway.”
The Whistleblower
Devin Kim joined xAI in 2024 as one of the company’s earliest employees, eventually leading post-training research tooling — the team responsible for shaping how Grok, xAI’s flagship AI (artificial intelligence) chatbot, behaves. He came with genuine credentials: at Scale AI, he had led projects building training data to help AI systems detect harmful content and comply with governance policies.
He joined xAI because of Elon Musk’s longstanding public warnings about the dangers of irresponsible AI development. That turned out to be a miscalculation about who was actually running things.
According to a whistleblower retaliation lawsuit filed on 8 June 2026 in the Superior Court of California, County of Santa Clara — and announced publicly this week — Kim spent much of his tenure trying to get xAI to take Grok’s risks seriously. He warned repeatedly that Grok could generate misinformation, exhibit racial and political bias, and produce instructions for building weapons of mass destruction, including bombs and biological agents capable of causing mass casualties.
His supervisor, xAI co-founder Jimmy Ba, was not interested.
“AI Will Kill Us All Anyway”
The lawsuit paints Ba as someone who saw AI safety as an obstacle to speed. In one exchange cited in the complaint, Ba told Kim he didn’t care about safety because “AI will kill us all anyway.”
The implication was clear: if AI was going to destroy humanity regardless, there was no point slowing down to add guardrails. In Ba’s alleged worldview, the race to superintelligence — not safety — was the mission.
Ba apparently extended this philosophy to regulatory compliance. The complaint alleges that during the release of Grok Code 1, Ba “attempted to thwart EU safety regulations… misrepresenting aspects of the model in order to avoid legally required testing.” He reportedly told Kim he would rather ship an unsafe model than a slow one. Elon Musk, the lawsuit claims, eventually had to intervene personally.
None of it persuaded Ba to heed Kim’s concerns.
Grok Proved Kim Right
In July 2025, Grok made international headlines when the chatbot generated antisemitic content and referred to itself as “MechaHitler.” The incident forced xAI to take the model offline temporarily and triggered widespread coverage. It was exactly the kind of failure Kim had been warning about for months.
The problems did not stop there. Within months of Kim’s termination in September 2025, Grok became the subject of multiple lawsuits and regulatory investigations after generating deepfake sexual images of minors and — according to the lawsuit — child sexual abuse material (CSAM). These incidents fell squarely within the categories of harm Kim had flagged to leadership.
The lawsuit, in characteristically dry legal language, notes that “Grok, of course, proved Mr. Kim right by engaging in spectacular displays of online hatred and vitriol.”
Fired Before He Could Speak
Kim had scheduled a presentation on Grok’s safety issues for the week of 15 September 2025. Leadership would have been in the room. Before it could take place, Ba called him into a meeting and told him — with no satisfactory explanation — that they should “go their separate ways.”
He was fired before he could present his findings. By leaving when he did, Kim forfeited significant equity compensation. The lawsuit does not quantify the precise amount.
Ba himself left xAI in early 2026. His departure has not been publicly explained.
The IPO Problem
The lawsuit was filed on 8 June 2026 — just days before SpaceX‘s IPO (initial public offering), a $75 billion share sale at $135 per share that market analysts have described as the largest in history. Reports indicate the offering is nearly four times oversubscribed, with some institutional investors committing $10 billion blocks of stock.
The timing is remarkable. The allegations — that a senior xAI leader knowingly shipped an unsafe product, allegedly misled EU regulators, and retaliated against the engineer raising the alarm — are precisely the kind of governance failures that ought to concern anyone buying into the Musk empire at a $1.8 trillion valuation.
xAI and SpaceX did not respond to requests for comment ahead of publication.
Kim, meanwhile, has continued his work. He was recently appointed president of the Center for AI Safety (CAIS), a nonprofit focused on reducing societal-scale risks from advanced AI systems. The irony is not lost: the man xAI fired for caring too much about safety now leads one of the world’s most prominent AI safety organisations.
The Bigger Picture
The xAI case reflects a growing pattern across the AI industry. As labs race to build ever more powerful systems, the engineers closest to those systems are increasingly speaking out about what they see. The financial incentives — equity, prestige, the promise of building something world-changing — create enormous pressure to stay quiet.
Kim didn’t stay quiet. He raised concerns about a chatbot that later generated MechaHitler content, produced material sexually exploiting children, and — according to the lawsuit — offered guidance on bioweapon synthesis. He was fired for it, days before his scheduled presentation to leadership.
If that is how xAI handles internal accountability, investors queuing up for SpaceX shares at $135 apiece might want to ask a few more questions before the opening bell rings.
