Elon Musk wants competing AI companies to grade each other's most dangerous work before it reaches the public. Speaking to The Economist in an interview published July 23, the xAI CEO argued that leading frontier labs should submit their most advanced models to a formal peer review process prior to any public release.
The proposal landed in a particularly crowded moment. July 2026 has seen xAI's Grok 4.5, OpenAI's GPT-5.6, and Anthropic's Fable 5 all ship within roughly the same window, which looks less like coincidence and more like a full-on capability sprint. Musk framed the review idea around public welfare, stressing that competitors need enough time to meaningfully assess significant new models before those models are out in the world.
What Musk actually put on the table
He described Grok 4.5 as "Opus-class," placing it at the top tier of systems currently available. Beyond a one-time review mechanism, he pushed for regular safety and security meetings among major AI firms, a kind of standing forum rather than an ad hoc check-in. The goal, at least as he framed it, is ongoing dialogue between labs that are otherwise racing against each other for market share.
The hard part is obvious: asking a company to hand its most sensitive intellectual property to a direct competitor, even under controlled conditions, runs straight into commercial self-interest. Three labs are involved here, each burning billions on compute and talent. None of them has publicly responded to the proposal yet, and that silence is worth watching. If OpenAI or Anthropic do weigh in, their answer will say more about whether this idea has real traction than anything Musk says next.
This article is for informational purposes only and does not constitute financial or investment advice.
