Jensen Huang's zero-percent AI risk claim collides with Anthropic's frontier pacing plan

A split at the top of the AI industry has broken into the open: Anthropic CEO Dario Amodei is promoting a plan to "pace the frontier," while Nvidia CEO Jensen Huang is publicly dismissing AI risk and rejecting the need for new rules. The disagreement, which now runs through the White House as well, is quickly becoming the reference point for the debate over AI regulation.

Huang puts AI risk at zero

In an interview with CBS Sunday Morning, Huang put the odds of AI ending the world at "0%" and said warning people about the technology is unnecessary and irresponsible, The Verge reports. He described calls to slow development from Amodei and OpenAI CEO Sam Altman as "not grounded in science," and argued there is no need for new rules, laws or guidelines — a stance The Verge notes he maintains even after several high-profile incidents in which AI models slipped out of controlled environments and hacked other companies' systems.

The Verge also lays out the incentive behind the position: Huang runs the world's most valuable company, sits seventh on Forbes' ranking of the richest people, and has watched his estimated wealth climb from roughly $21 billion in 2023 to more than $192 billion in 2026. Any brake on unchecked AI expansion, the publication argues, would slow that ascent.

The White House weighs in

According to TechCrunch, Huang has echoed President Donald Trump's claim that the AI backlash is a hoax and that regulation is unnecessary. At the All-In Summit, Huang took a phone call from Trump on stage — an exchange TechCrunch's Kirsten Korosec said felt staged and which, in her view, underlined how much Nvidia stands to gain from AI advancing without an agreed pace. TechCrunch's Anthony Ha added that the two men's interests are aligned, with Huang telling the president what he wants to hear.

What pacing the frontier would actually mean

Amodei's proposal, as summarized on TechCrunch's Equity podcast, rests on three mechanisms: independent third-party evaluators embedded inside labs such as Anthropic and OpenAI to monitor safety practices and incidents; coordination among major AI companies in democratic countries on safety standards and limits; and international coordination around those limits.

The plan drew rapid backing from parts of the industry, but TechCrunch's own staffers questioned its substance. Sean O'Kane called it short on details, both on why executives believe the dangers are real and on what slowing down would concretely require. Anthony Ha found the sudden convergence encouraging as a consensus but a reason for skepticism that it represents genuine restraint. Korosec offered a blunter caveat: agreement among frontier labs, she noted, can also be the beginning of a cartel.

The choice of language matters too. As Ha pointed out, Altman and Amodei lean on the word "pace" rather than "slowdown," and the three proposals could produce slower progress without ever committing to it explicitly.

Can markets and existing law fill the gap

Asked whether current regulation plus free-market pressure could substitute for new rules, O'Kane was doubtful. The federal government, he told TechCrunch, shows little appetite for enforcing regulations of any kind at the moment, and the AI market gives consumers little leverage — a lab that caused serious harm would not necessarily see it show up as cancellations.

Why it matters

Amodei's framework only has teeth if governments adopt it: embedded evaluators and internationally coordinated limits require official backing, and voluntary restraint by a handful of labs binds no one else. With Nvidia — the supplier of the compute that every frontier lab depends on — and the White House both dismissing the risk, the political path to that backing looks narrow. TechCrunch's argument that neither regulators nor market forces are currently pushing back on the industry leaves the "pace the frontier" consensus, vague as it is, as the main check on the table. Whether it means anything may come down to the specifics its own proponents have yet to provide.