Musk Proposes Peer-Review Safety Testing for U.S. and Chinese AI

Elon Musk has proposed a voluntary peer-review framework for AI safety. He calls for major U.S. and Chinese labs to allow competitors to run safety test harnesses on new models before public release. The proposal shifts industry practice away from self-regulation toward a model where rivals evaluate each other's work to identify potential hazards.

"So, you know, instead of grading your own homework, you would at least have competitors grading your homework and raising the alarm if they see concerns," said Elon Musk, CEO of SpaceX, according to a report by CNBC.

The Scope of the Proposal

Musk's initiative targets the leading edge of global artificial intelligence development. He suggested that his own firm, xAI, along with OpenAI, Anthropic, Google, and Meta, should open their proprietary systems to inspection by one another. He also recommended including top Chinese AI companies in this testing effort.

While the proposal aims to increase transparency, Musk acknowledged that no rival AI laboratories have agreed to participate. The suggestion comes as the industry faces pressure to address the risks posed by frontier models.

Industry Concerns Over AI Alignment

The debate over AI safety has intensified because development speeds outpace existing regulatory frameworks. Some researchers within the industry have expressed alarm regarding the trajectory of these technologies. Jacob Coxon, who previously worked at both Anthropic and OpenAI, has characterized the current development environment as "gambling with our lives."

This sentiment is echoed by other senior figures in the field. Evan Hubinger, an alignment lead at Anthropic, noted the severity of these internal concerns, stating, "Jacob is correct here. We really do earnestly believe AI could kill all humans. I personally think it is >10% within the next decade."

Context: The Regulatory Vacuum

Musk’s call for voluntary peer review arrives at a time when formal government regulation of AI remains largely stalled. The current administration has signaled a rejection of heavy-handed federal oversight, preferring to rely on market-led safety standards. By advocating for a peer-review mechanism, Musk is attempting to establish a private-sector alternative to federal mandates.

This approach differs from historical safety protocols in industries like aviation or pharmaceuticals, where independent regulatory bodies perform mandatory testing. In the absence of such oversight for AI, the proposal highlights a tension between the need for rigorous safety verification and the competitive desire to keep proprietary model architectures secret.