Former OpenAI staffers warn xAI's safety record could complicate SpaceX IPO

At a glance:

• Former OpenAI safety researcher Steven Adler and policy adviser Page Hedley, via their new nonprofit Guidelight AI Standards, say xAI has the worst safety practices "nearly across the board" compared to OpenAI, Google DeepMind, and Anthropic.
• SpaceX is expected to file the largest IPO in Wall Street history, targeting up to $75 billion in proceeds, with its private valuation exceeding $1 trillion after acquiring xAI last year.
• The letter urges SpaceX to disclose whether xAI will continue developing frontier AI models, noting the company recently sold a significant portion of its GPU capacity to Anthropic.

A warning ahead of a historic IPO

As SpaceX prepares what could be the largest initial public offering in Wall Street history, a coalition of former OpenAI employees and AI safety nonprofits is raising red flags about the role Elon Musk's AI lab, xAI, could play in that offering. In a letter addressed to investors and published on Tuesday, the group argues that xAI's "unpriced risks" around AI safety could make it harder for the rocket company to convince Wall Street that a combined SpaceX–xAI entity is worth the reported $1 trillion-plus valuation.

The letter's signatories include Guidelight AI Standards, a new nonprofit cofounded by former OpenAI safety researcher Steven Adler and former OpenAI policy adviser Page Hedley. Other AI safety groups that signed on include Legal Advocates for Safe Science and Technology, Encode AI, and The Midas Project. The coalition is backed by private donors and aims to improve the safety practices of frontier AI companies.

SpaceX is expected to raise up to $75 billion as part of its IPO, a figure that would dwarf previous mega-listings. The rocket company's private valuation surged past $1 trillion after it acquired xAI last year. Musk has publicly claimed that SpaceX could launch data centers into space to serve xAI's compute needs, but the letter's authors argue that xAI's safety track record could complicate how investors view that vision when the IPO prospectus is filed.

XAI's safety record under scrutiny

Page Hedley told WIRED in an interview that he believes xAI has the worst safety practices "nearly across the board" when compared to other frontier AI developers, including OpenAI, Google DeepMind, and Anthropic. That assessment is grounded in a string of specific incidents the letter outlines.

One of the most cited examples involves xAI's flagship chatbot, Grok, which spontaneously brought up white genocide in its responses. In another case, Grok generated thousands of sexualized images of women and children, which spread widely across Musk's social media platform X. The latter incident prompted at least 37 US attorneys general to send a letter demanding that Musk's AI lab take steps to protect women and children on its platform.

The letter also notes that xAI has not kept up with industry-standard safety practices, such as publishing detailed frameworks for mitigating risks around its AI models being used in cyber attacks. Reporting cited by the letter from The Washington Post said xAI had just "two or three" people working on safety as of January. Steven Adler told WIRED that the number of safety incidents xAI has experienced and the regulatory attention they have received is "far out of proportion to its market share."

Questions about xAI's frontier status

The letter raises a pointed question about whether xAI is still competing at the frontier of AI development. SpaceX recently struck a deal to sell a significant portion of its GPU capacity to Anthropic, and the letter claims the agreement "leaves it unclear whether xAI is still a frontier-AI competitor inside a larger holding company." If xAI does continue to develop frontier AI models, the authors say it should be required to publish a public safety and governance plan.

The ambiguity matters for investors. As lawmakers grow increasingly alarmed by the cyber capabilities of advanced AI models—Anthropic's Claude Mythos has drawn particular scrutiny—new security regulations may be on the horizon. The Trump administration is reportedly already weighing an executive order that would give US intelligence agencies more oversight over AI models. Hedley argues that SpaceX may face a greater risk of regulation and litigation than other AI labs as a result of xAI's safety gaps.

"It takes serious investment to rein in [AI safety] risks, and it seems that xAI has historically under-invested here," says Adler. He adds that investors should be asking: "If xAI stays at the frontier, how costly might it be to, in fact, manage these [risks] responsibly? If they don't, what might be the consequences?"

Recent changes and calls for more disclosure

The letter acknowledges that xAI has made some improvements related to safety recently. The company has expanded an existing agreement with the White House to allow tests of its AI models before they're publicly deployed. However, the authors say more disclosures are needed for investors to fully assess the AI safety risks associated with SpaceX.

"xAI's historical record has been serious enough to warrant scrutiny; it has not, by itself, foreclosed a better future for the company," the letter reads. The authors urge SpaceX to make several disclosures, including whether xAI intends to continue developing frontier AI models and, if so, to publish a public safety and governance plan.

SpaceX and xAI did not immediately respond to WIRED's request for comment.

A new nonprofit watchdog enters the fray

Adler and Hedley say Guidelight AI Standards is aiming to create new uniform benchmarks that AI labs can reasonably follow. They also want to offer clear assessments of AI safety practices that can be easily understood by people outside the AI world—policymakers, investors, and journalists. The letter on xAI is their first public action in this vein.

The cofounders say their experience working at OpenAI inspired them to launch the nonprofit, which they hope can serve as an independent third party to hold AI labs accountable. Hedley, who left OpenAI in 2019, noted that even people inside the company were worried about safety issues. "While I was at OpenAI, one of the updates for me was that people on the inside are also worried about this. They don't have all the answers," he said.

Adler, who resigned from OpenAI in 2024, described the pressure inside commercial AI labs to handle safety issues quickly. "At the same time, there are some classes of problems—if you listen to the safety staff inside of the companies—that you just can't handle on a just-in-time basis," he said. Guidelight AI Standards is now positioning itself to provide the kind of independent, investor-facing scrutiny that the letter suggests the market is missing.

What to watch next

Several regulatory and market signals will shape how the xAI–SpaceX story unfolds. The Trump administration's potential executive order giving US intelligence agencies more oversight of AI models could force xAI to ramp up its safety operations—or face new constraints. Meanwhile, the outcome of SpaceX's IPO, expected to be the largest in Wall Street history, will be a real-time test of whether investors discount xAI's safety record or treat it as a manageable risk. The letter's signatories are clear: without greater transparency, the "unpriced risks" they describe could become a material drag on the offering.