Bessent rejects liability exemptions for AI labs as Huang says safety needs no new laws
The Treasury secretary told House lawmakers creators should be liable for what they build, while Nvidia's CEO called safety an engineering problem and Cohere's Aidan Gomez warned against Silicon Valley self-regulation.
Treasury Secretary Scott Bessent told the House Financial Services Committee on Tuesday that the government should not grant frontier AI labs the liability exemption he says they are asking for. Answering Rep. Juan Vargas, Bessent said Treasury has been working on safety nonstop since the release of Anthropic's Mythos model, whose cybersecurity risks prompted an April meeting at Treasury with then-Fed Chair Jerome Powell and bank CEOs. The best way to guarantee safety, he said, is to hold AI creators liable for what they build and generate. The same day, Nvidia CEO Jensen Huang told Salesforce's Dreamforce audience that no new laws or regulations are needed. Safety is an engineering problem, not a legal one, he said, and market forces already push companies not to release products they are not confident in. Huang rejected the description of AI as an alien mind, calling it a complicated but ordinary computing system built by humans, and said speed and safe products are not a trade-off. In Toronto, Cohere CEO Aidan Gomez told a national investment summit that a small group of Silicon Valley companies should not be left in charge of AI safety. He said a CEO asking for antitrust laws, which were written to prevent cartels, to be relaxed for their own company should raise a red flag. The three statements show the policy debate splitting along liability, self-regulation, and competition lines rather than converging on a slowdown.