- xAI founder proposes rival AI companies test and grade one another’s models before public release
Elon Musk is calling on competing artificial intelligence companies to test each other’s AI models for safety before releasing them to the public, arguing that rival evaluations could identify problems that internal testing misses.
Musk made the proposal during the All-In Summit in Los Angeles, where he said leading AI companies should allow competitors to evaluate their systems rather than relying solely on internal safety reviews. Fox News reported that Musk said the approach could be implemented immediately.
What Elon Musk Proposed for AI Safety
Musk's proposal would effectively have competing AI labs test and grade one another's models before release. He compared the concept to having someone other than the developer check its work.
The idea is intended to create another layer of scrutiny around increasingly capable AI systems. Musk said companies could face significant legal and reputational consequences if competitors identified safety problems before release and those warnings were ignored.
Musk also said any broader AI safety arrangement would need to be acceptable to China, reflecting concerns about the competitive relationship between the US and China in artificial intelligence development.
Why AI Companies Are Debating Model Testing
Musk's comments come as AI executives debate how companies should evaluate increasingly powerful models and AI agents. Anthropic CEO Dario Amodei has called for a slower pace of frontier AI development, while OpenAI CEO Sam Altman has backed efforts to pace AI progress. Other industry leaders have discussed independent evaluations and additional safeguards for advanced AI systems.
Meta Platforms Inc. (META) CEO Mark Zuckerberg has taken a different approach, arguing that individual AI companies have the responsibility and incentive to develop their models safely. Zuckerberg has nevertheless pointed to independent evaluators as an existing industry practice.
The discussion has also expanded into proposed government oversight. Bipartisan lawmakers have discussed legislation that would require major AI developers to publish safety plans, report serious incidents and undergo independent evaluations.
Could Rival AI Companies Test Each Other?
Musk's proposal would require competing AI developers to agree on testing standards, access to models and procedures for reporting safety concerns. There are also practical questions about how companies could conduct cross-testing without exposing confidential technology or other commercially sensitive information.
For now, Musk's proposal remains a suggested approach rather than an established industry standard. The broader debate over AI safety continues to involve company-led testing, independent evaluations, and potential government oversight.
What Musk's AI Safety Proposal Means
The proposal highlights a central issue in AI safety: whether companies should rely primarily on internal testing or introduce greater outside scrutiny before releasing advanced models.
Rival testing could provide another avenue for identifying potential weaknesses, but implementing such a system would require cooperation among companies that are competing to develop increasingly capable AI technology.