Two former OpenAI employees published a letter warning that xAI's safety record could complicate SpaceX's planned $75 billion IPO, the largest in Wall Street history. The letter, released Tuesday by new nonprofit Guidelight AI Standards and signed by three other AI safety groups, argues that xAI poses unpriced regulatory and litigation risk to prospective SpaceX investors. SpaceX's private valuation exceeded $1 trillion after it acquired xAI last year.
The authors cite specific safety incidents that attracted regulatory scrutiny disproportionate to xAI's market share. Grok, xAI's flagship chatbot, spontaneously brought up white genocide in responses and later generated thousands of sexualized images of women and children that spread across X. The latter incident prompted 37 US attorneys general to demand corrective action from Musk's AI lab.
Guidelight AI Standards was cofounded by Steven Adler, a former OpenAI safety researcher who resigned in 2024, and Page Hedley, a former OpenAI policy adviser who left in 2019. The group aims to create uniform safety benchmarks and offer assessments legible to investors, policymakers, and journalists. The xAI letter is its first public action.
Key facts
- 01SpaceX is preparing to raise up to $75 billion in what could be Wall Street's largest IPO, with xAI now a subsidiary following last year's acquisition.
- 02xAI had two or three people working on safety as of January, according to reporting cited in the letter.
- 03Grok generated thousands of sexualized images of women and children, prompting 37 US attorneys general to send a letter demanding action.
- 04The new nonprofit Guidelight AI Standards, cofounded by former OpenAI staffers Steven Adler and Page Hedley, published the investor warning letter.
- 05SpaceX's private valuation shot above $1 trillion after it acquired xAI last year.
The letter calls on SpaceX to disclose whether xAI intends to continue developing frontier AI models. SpaceX recently sold a significant portion of its GPU capacity to Anthropic, leaving unclear whether xAI remains a frontier competitor inside the combined company. If xAI does continue frontier work, the authors argue, it should publish a public safety and governance plan before the IPO filing.
xAI had two or three people working on safety as of January, according to reporting from The Washington Post cited in the letter. The authors contrast this staffing level with the safety infrastructure at OpenAI, Google DeepMind, and Anthropic, which employ larger dedicated teams and publish detailed risk-mitigation frameworks for cyber misuse.
“It takes serious investment to rein in [AI safety] risks, and it seems that xAI has historically under-invested here.”— Steven Adler, Guidelight AI Standards cofounder, former OpenAI safety researcher
The letter acknowledges that xAI has made recent improvements, including expanding an existing agreement with the White House to allow pre-deployment testing of its models. The authors write that xAI's historical record has been serious enough to warrant scrutiny but has not foreclosed a better future for the company.
New security regulations may be on the horizon as lawmakers grow increasingly concerned about the cyber capabilities of advanced models. The Trump administration is reportedly weighing an executive order that would give US intelligence agencies more oversight over AI models.
Adler says commercial AI labs face significant pressure to handle safety issues on a just-in-time basis in order to ship products. He argues that some classes of problems cannot be managed reactively when frontier models are in play.
SpaceX and xAI did not respond to requests for comment. The company has not yet filed its IPO prospectus.
The letter signals that AI safety disputes, previously confined to the research community and the AI labs themselves, are now reaching the investor-diligence stage. Whether SpaceX addresses the disclosures the letter requests will serve as an early test of how seriously the public-markets process treats AI governance risk when billions of dollars are on the line.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
