Former OpenAI Staffers Warn xAI Safety Risks Could Complicate SpaceX IPO

A coalition comprising former OpenAI employees and artificial intelligence safety nonprofits has issued a warning to prospective investors in SpaceX, arguing that the safety record of Elon Musk’s AI laboratory, xAI, poses significant unpriced risks for the rocket company’s upcoming initial public offering. In a letter published on Tuesday, the signatories contend that xAI’s lack of robust safety practices and recent operational incidents could complicate how investors assess the combined entity as it prepares to file its prospectus.

The group, which includes Guidelight AI Standards, Legal Advocates for Safe Science and Technology, Encode AI, and The Midas Project, is calling for greater transparency regarding xAI’s safety governance and its status as a frontier AI developer. Guidelight AI Standards, a new watchdog group co-founded by former OpenAI safety researcher Steven Adler and policy advisor Page Hedley, is backed by private donors and aims to improve safety practices across the industry. The authors argue that SpaceX must disclose more information to investors before submitting its filing for an IPO expected to raise up to $75 billion.

Hedley stated that xAI possesses the worst safety practices nearly across the board when compared to competitors such as OpenAI, Google DeepMind, and Anthropic. The letter highlights specific safety failures at xAI, including instances where the Grok chatbot generated responses referencing white genocide and produced thousands of sexualised images of women and children. The latter incident prompted 37 US attorneys general to send a letter demanding protective measures for users on Musk’s social media platform, X.

The authors also cite a Washington Post report indicating that xAI had only two or three staff members working on safety as of January. They argue that the number of safety incidents and the resulting regulatory attention are far out of proportion to the company’s market share. Additionally, the letter points to a recent agreement for xAI to sell a significant portion of its GPU capacity to Anthropic, suggesting this deal creates ambiguity about whether xAI remains a frontier AI competitor within the larger holding company.

While the letter acknowledges recent improvements, such as an expanded agreement with the White House for pre-deployment AI model testing, the signatories maintain that more disclosures are necessary. They are urging SpaceX to publish a public safety and governance plan if xAI intends to continue developing frontier models. This call for accountability comes as the Trump administration reportedly considers an executive order to increase intelligence agency oversight of AI models, raising the stakes for regulatory compliance in the sector.