We don’t need AI regulation — leave safety to us, Nvidia’s Jensen Huang says

Julie Bort at TechCrunch reports Nvidia CEO Jensen Huang telling Salesforce’s Dreamforce audience that AI isn’t an “alien mind” but hardware and software built by humans—therefore controllable with existing approaches. “Safety is an engineering problem, not a legal one,” he said. If you’re not confident in safety, don’t ship; markets will punish those who do.

That stance collides with a week of frontier-lab letters pleading for pacing, evaluators, and even antitrust waivers for safety talks.

The chipmaker that powers the race is telling Washington: don’t add new rules.

We’re not buying “engineering only” as a complete civic answer.

Huang’s line is coherent from a systems vendor’s chair: treat AI like complicated computing, reuse product-liability instincts, let non-confident teams withhold release. There’s a useful kernel—many harms are ordinary software harms, and shipping discipline matters.

But market punishment arrives after damage. Existing law didn’t stop unsafe cars without regulation layered on; “don’t release it” failed as the sole check in other high-stakes industries. When OpenAI researchers talk alien minds and Anthropic’s CEO asks for embedded evaluators, Huang’s dismissal reads as strategic, not merely philosophical.

We’re for useful AI that ships under verification—evaluators, incident reports, liability with teeth—not a veto on progress. The harm of Huang’s frame at scale: policy delay while capability and deployment compound.

I’m watching whether Nvidia’s “no new laws” line shapes U.S. drafts this session, and how Huang answers catastrophic-risk scenarios beyond ordinary product defects.

Huang spoke at Salesforce Dreamforce on September 15, 2026. The “alien mind” contrast references recent OpenAI safety-researcher language reported in the broader press cycle.