Former xAI employee Devin Kim filed a lawsuit against the company and its parent company SpaceX, claiming that he was fired for voicing concerns about artificial intelligence safety. The lawsuit, filed in California state court on Tuesday, came to the fore at a time when expectations regarding SpaceX’s public offering process were intensifying. According to the lawsuit, Kim argued that security measures were not taken into account sufficiently in the development of the artificial intelligence chat bot called Grok, while he was working at xAI, and repeatedly warned the management about this issue.
In the lawsuit reviewed by TechCrunch, it is stated that Kim is particularly concerned about the possibility that Grok may encourage discrimination and contribute to the spread of information about weapons of mass destruction. While it is reminded in the lawsuit that Grok has been criticized over time for various security and behavioral problems, it is claimed that the model’s association with hate speech and extreme content at some periods confirms Kim’s warnings. The documents also state that Kim carried out efforts to re-evaluate the model’s political prejudices and discriminatory tendencies after Grok used expressions that compared him to Hitler at one point.
A few months after Kim left the company, Grok was at the center of another controversy. Allegations that the chat bot was used to spread non-consensual sexual images on the X platform, which is owned by Elon Musk and operates under the umbrella of xAI, have received wide coverage. These developments have brought the debates about the security of artificial intelligence systems and content control back to the agenda.
The lawsuit also points to disagreements in xAI management
The lawsuit positions Kim not only as an employee but also as a whistleblower who warns against practices within the company. The documents include assessments that xAI’s approach to artificial intelligence security may have unlawful consequences in terms of internet regulations, consumer protection rules, unfair commercial practices and weapons-related legislation. Despite this, the lawsuit does not directly target Elon Musk. On the contrary, Kim’s lawyers claim that Musk gave instructions to comply with legal obligations and carry out the necessary security tests.
The focus of the allegations is Jimmy Ba, one of the co-founders of xAI. It is claimed that Ba, who left the company this year, opposed Kim’s demands for security measures and tried to silence the employee. According to the lawsuit, Ba rejected various suggestions regarding AI safety and took retaliatory actions against Kim’s repeated complaints. The documents also claim that Ba said during a meeting, “Artificial intelligence will kill us all anyway.”
Another noteworthy claim in the case file belongs to the period August 2025. Kim’s lawyers claim that Ba tried to avoid mandatory testing processes by misrepresenting some features while the Grok Code 1 version was being evaluated under European Union regulations. It is stated in the documents that Ba preferred to release a model that looked more powerful, even if it was risky in terms of security, rather than a low-performance model, and Musk had to intervene in the process.
It is reported that Kim planned to present his findings within the company in mid-September 2025, but was called to a meeting by Ba before this presentation took place. According to the case file, it is stated that in this meeting, Kim was asked to part ways with the company, but he was not given a satisfactory reason.
Kim, whose work in the field of artificial intelligence security dates back to xAI, previously took part in projects at Scale AI that developed training data on detection of harmful content and compliance with governance policies. Last week, he was appointed president of the Center for AI Safety, a non-profit organization focusing on artificial intelligence risks. In addition to demands for compensation and criminal sanctions, the lawsuit filed by Kim also demands that the court determine that xAI and SpaceX’s behavior is unlawful.