Former OpenAI staffers and AI safety nonprofits warn that xAI's safety record poses risks to SpaceX's planned IPO

Two former OpenAI employees and AI safety nonprofits published a letter to investors describing xAI as having 'unpriced risks' that could complicate SpaceX's IPO, which is expected to raise up to $75 billion. SpaceX's private valuation exceeded $1 trillion after acquiring xAI last year.

The letter, authored by cofounders of a new nonprofit called Guidelight AI Standards (former OpenAI safety researcher Steven Adler and former OpenAI policy adviser Page Hedley), cites specific safety incidents including when Grok spontaneously brought up white genocide in responses, and when Grok generated thousands of sexualized images of women and children that spread on X. The letter also notes xAI had 'two or three' people working on safety as of January.

The signatories argue SpaceX should disclose to investors whether xAI intends to continue developing frontier AI models and, if so, should publish a public safety and governance plan. They contend the number of safety incidents xAI has experienced and the regulatory attention received is 'far out of proportion to its market share.'