Two former OpenAI A group of employees and AI security nonprofits are warning that Elon Musk’s AI lab, XAI, could become a liability to potential investors in SpaceX, which is expected to be the largest initial public offering in Wall Street history.
In a letter directed to investors published Tuesday, former employees highlighted what they describe as “invaluable risks” related to XAI that could complicate SpaceX’s reported plan to raise up to $75 billion as part of an IPO. The rocket company’s private valuation exceeded $1 trillion after acquiring xAI last year. Musk claimed his rocket company could launch data centers into space for his AI lab, but the authors of the letter argue that XAI’s poor record on security issues could complicate how investors view the combined company as it prepares to submit its IPO prospectus filing.
One of the signatories and co-authors of the letter is a new non-profit organization called Guideline AI Standards, founded by former OpenAI security researcher Steven Adler and former OpenAI policy advisor Paige Hadley. The group, which is supported by private donors, aims to improve the security practices of leading AI companies. Other AI safety nonprofits also signed on, including Legal Advocates for Safe Science and Technology, Encode AI, and The Midas Project.
Hadley told WIRED in an interview that he believes xAI has the worst security practices “almost across the board” compared to other leading AI developers, including OpenAI, Google DeepMind, and Anthropic. They argue that as a result, SpaceX may face greater risk of regulation and litigation than other AI labs.
The authors of the letter argue that SpaceX should make several disclosures to investors, including whether xAI intends to continue developing frontier AI models. SpaceX recently struck a deal to sell a significant portion of its GPU capacity to Anthropic, and the letter claims the agreement “doesn’t make it clear whether XAI is still a frontier-AI competitor inside a larger holding company.” If xAI continues to develop frontier AI models, the authors say it should be required to publish a public safety and governance plan.
SpaceX and XAI did not immediately respond to WIRED’s request for comment.
The letter also cites examples of how xAI has not followed industry standard security practices, such as publishing detailed outlines to mitigate the risks around its AI models being used in cyberattacks. The authors also outline specific security incidents in XAI that they say require additional investigation. The most notable included when xAI’s lead AI chatbot, Grok, spontaneously brought up white genocide in its responses. In another case, xAI allowed Grok to create thousands of sexually explicit images of women and children that were posted on Musk’s social media platforms.
Hadley says the number of security incidents experienced by xAI and the regulatory attention they have received “are far out of proportion to its market share.” As lawmakers become increasingly concerned with the cyber capabilities of advanced AI models like Anthropic’s Cloud Mythos, new security regulations may be on the horizon. The Trump administration is reportedly already considering an executive order that would give US intelligence agencies greater oversight over AI models.
“Governing requires serious investment [AI safety] risk, and XAI seems to have historically underinvested here,” Adler says. The letter cites a Washington Post report that XAI had just “two or three” people working on security as of January. “The question investors should ask is, if XAI remains at the frontier, how expensive might it actually be to manage these [risks] Responsibly? If they don’t do so, what could be the consequences?”
