Nvidia's chief executive, Jensen Huang, has stated that ensuring artificial intelligence safety does not require new legislation, arguing that existing market mechanisms provide sufficient checks and balances.
Speaking at the Salesforce Dreamforce conference in San Francisco on September 15, Huang emphasized during a conversation with Salesforce CEO Marc Benioff that companies should set their own product release timelines, ensuring they only launch offerings that can gain market acceptance. His remarks directly address the widespread debate over AI governance, carrying significant weight coming from the leader of the world's largest AI chip supplier as regulators and lawmakers discuss how to oversee the technology.
Innovation and safety are not mutually exclusive — market forces replace regulation
Huang explicitly rejected the idea that AI innovation speed and product safety are opposing priorities, calling it a "false dichotomy." He stated that both objectives can absolutely be achieved simultaneously. He believes companies can fully prioritize safety while advancing AI development, with the key being careful management of release timing until there is strong confidence in market readiness.
On the governance front, Huang favors relying on market dynamics rather than government intervention. He noted that market forces have long existed and are sufficient to provide effective constraints on AI product safety. This position stands in clear contrast to some policymakers' proposals, and as discussions intensify across major economies over whether AI requires dedicated legislation, Huang's stance adds a new dimension to the debate.
Federal legislative direction focuses on extreme risks and national security
The U.S. Congress currently lacks a unified federal law comparable to the European Union's AI Act, instead pursuing targeted proposals centered on extreme risk prevention and national security. Bipartisan lawmakers recently introduced the "AI Kill Switch Act," which would require developers of the most advanced AI systems to maintain the capability to limit, suspend, or completely shut down their systems; if a system shows signs of uncontrolled behavior that could lead to catastrophic harm, the Department of Homeland Security would have the authority to order its deactivation.
Meanwhile, a group of senators led by Bernie Sanders is pushing legislation to pause the development and deployment of "superintelligent AI" until federal safety review rules are finalized. On national security, the Advanced AI Security Readiness Act is under congressional review, requiring the National Security Agency to develop defense plans against foreign actors stealing core AI technologies. The Senate Armed Services Committee has also included provisions in the annual defense authorization bill mandating human oversight in military AI weapons use, with fully autonomous decision-making prohibited.
At the executive level, the White House signed an executive order in June 2026 promoting advanced AI innovation and safety, and established an AI cybersecurity clearinghouse to address vulnerabilities in frontier models.
Comments