Nvidia CEO Says We Don't Need New AI Laws

Nvidia founder and CEO Jensen Huang used his talk at Salesforce's Dreamforce conference on Tuesday, September 15, to argue against new AI laws. AI safety is "an engineering problem, not a legal one," he said, according to TechCrunch.

He said AI is hardware and software made by people, so people can control it with laws we already have. He said we do not need separate rules for building, releasing or using AI. Rules should focus on clear, measurable harms, not broad limits on research or products. He gave the same view in early September, according to Yahoo Finance.

He said the job falls on sellers. Companies should not release a product unless they trust how it works and that it is safe. They should slow down or wait until it is right. He said the market will handle the rest because buyers will not buy unsafe products.

He also rejected calling AI an "alien mind." He said AI systems are built things, like cars. They can be tested, fixed and called back if something is wrong.

His comments came as leaders debated AI rules. In early September, the United States asked G20 members to avoid new AI rules and leave the technology alone, according to Reuters. President Donald Trump said in September the United States already has guardrails to control and punish AI companies, according to Reuters. On September 14, TechCrunch reported Huang had told Trump "we're not going to let [an AI slowdown] happen," in a piece by Amanda Silberling.

Not every lab agrees with leaving AI alone. In August 2026, OpenAI asked California to add more safety steps to its AI safety bill from the prior year.

Nvidia's own statements are more careful than a no-new-laws headline sounds. Its AI trust center says AI should respect privacy and data rules, stay safe and secure, and work in a clear way where people can be held responsible. In its Open Weights and American AI Leadership paper in July, Nvidia said sharing models openly may be one of the most important ways to keep AI safe and secure, and keeping models closed is not automatically safe. The company met with White House, Congress and tech leaders in September 2023 to talk about AI standards and best practices.

Huang often calls AI basic infrastructure, not just one app. He said AI is no longer a single breakthrough, it is infrastructure every company will use and every nation will build. At Dell Technologies World, he called every company an "Intelligence Manufacturer." In March 2026, he said every company should have an OpenClaw strategy. In April 2025, he appeared to have made a deal with the Trump administration to avoid limits on exports of Nvidia's H20 AI chips.

The broader context here helps explain why a chip seller favors engineering answers. For builders, safety as engineering means everyday safety work: standard tests, trying to trick the system to find flaws, tracking training data, limiting access, watching the system while people use it, responding when something breaks, and releasing updates in small steps.

In my view, shoppers handle clear failures well and hidden failures poorly. People will drop a chatbot that invents facts or a helper that deletes files. They are slower to spot unfair scoring, quiet leaks of private data, or shared breakdowns when many companies use the same base models and safety tools. Careful building is needed. Whether it can replace outside rules depends on what buyers can see before they buy.

Looking at what this means for builders, sharing models openly may matter more than fighting new laws. If wide review, repeatable tests and quick fixes make AI safer, then export rules, license terms and rules about what companies must share matter as much as any safety bill. Roads and power lines are controlled through buying rules, insurance and courts before new laws come. AI may follow that path. That is where Huang's view will be tested, and where good engineering can help businesses trust AI with more real work.