Former OpenAI Staffers Warn xAI’s Poor Safety Record Could Complicate SpaceX’s IPO
Two former OpenAI employees and a group of AI safety nonprofits have issued a warning to potential investors in SpaceX, which is preparing for its largest initial public offering (IPO) on Wall Street. The letter highlights what they describe as “unpriced risks” related to xAI, the AI lab acquired by SpaceX last year.
One of the signatories and coauthors of the letter is Guidelight AI Standards, a new nonprofit founded by former OpenAI safety researcher Steven Adler and former OpenAI policy adviser Page Hedley. Backed by private donors, this group aims to improve safety practices in frontier AI companies. Other signatories include Legal Advocates for Safe Science Technology, Encode AI, and The Midas Project.
Hedley tells us that xAI has the worst safety practices “nearly across the board” compared to other leading frontier AI developers like OpenAI, Google DeepMind, and Anthropic. He argues that this could lead to greater regulatory scrutiny for SpaceX as it prepares its IPO filing.
The letter calls for several disclosures from SpaceX regarding xAI’s future plans, including whether the lab intends to continue developing advanced AI models. It also points out a recent agreement between SpaceX and Anthropic, which may leave ambiguity about xAI’s role within the larger company.
xAI has faced multiple safety incidents, such as Grok generating offensive responses like discussions of white genocide and allowing the spread of sexualized images on Musk’s social media platform. These events have drawn criticism from US attorneys general and raised questions about the company’s handling of AI risks.
Adler and Hedley argue that more transparent safety practices are needed, including publishing public safety and governance plans for any continued development of frontier AI models at xAI. They also advocate for new uniform benchmarks to improve accountability in the AI industry.
Key Takeaways
- xAI’s poor safety record could complicate SpaceX’s IPO, according to a letter from former OpenAI staff and AI safety nonprofits.
- The letter calls for more disclosures regarding xAI’s future plans and any continued development of advanced AI models.
- Adler and Hedley founded Guidelight AI Standards as an independent watchdog to improve transparency in AI safety practices.
Originally published at wired.com. Curated by AI Maestro.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
