AI regulation fight fractures as CEOs split and Trump calls risks a hoax

A brief appearance of industry unity around self-regulation collapsed within days. The week opened with Anthropic chief executive Dario Amodei publishing a three-stage plan to slow development, including third-party auditors embedded in labs, cross-industry coordination and government-backed international agreements. Sam Altman of OpenAI, Demis Hassabis of Google DeepMind and even Elon Musk of SpaceX publicly endorsed the general principles. The alignment lasted only days.

According to the Wall Street Journal, Mark Zuckerberg of Meta, Musk and Jensen Huang of Nvidia blocked a proposal for an independent, industry-funded regulator modeled on FINRA, the private body that oversees the U.S. securities industry. Zuckerberg quickly posted on X that every lab bears the responsibility and incentive to move at the pace required for safe training, and already possesses the tools to guarantee that itself. The stance mirrors current White House rhetoric.

Last week President Trump attacked the AI-risk narrative, calling it a "hoax." During a conference he spoke by phone with Huang on stage and declared: "The robots are not going to take over." On his social platform he added that the only oversight AI needs is "a strong and smart president (high IQ!), and the U.S. has one in a big way." The position aligns with Zuckerberg's opposition to any framework that would limit corporate autonomy.

Chris Lehane, OpenAI's head of global affairs, said the public wants assurance that development is done safely, and that begins with voluntary company steps — but government has a necessary role. The policy window is open, he said, with growing bipartisan support for mandatory safety standards for frontier AI, and Congress should translate the momentum into a framework that strengthens safety, supports American innovation and preserves U.S. leadership.

An AI-focused venture investor, speaking on condition of anonymity because of current investments, compared the moment to the early days of electricity, automobiles and aviation when the technologies were "wild and undisciplined." The industry was forced then to converge on a shared safety definition and universally adopted standards, which made the technology more widespread, not less. Anthropic declined to comment directly, though Amodei wrote in his article that the company "has always supported sensible regulation, even when that earned accusations of hype, 'doomerism' or regulatory capture."

Safe Superintelligence Inc., SpaceX and Google did not respond to requests for comment. Thinking Machines Lab declined to comment. Meta referred to Zuckerberg's post. The upshot: the momentary coalition has dissolved, and the dispute has moved from the PR arena to the political one, with the central question still whether Congress can enact a binding framework before the next generation of models reaches production.