Nvidia CEO Jensen Huang has weighed in on the ongoing debate surrounding artificial intelligence safety, advocating for a company-led approach rather than new governmental regulations. Speaking at Salesforce’s Dreamforce conference, Huang characterized AI safety as fundamentally an engineering problem that companies are responsible for solving.
“Safety is an engineering problem, not a legal one,” Huang stated. He emphasized that AI, being a combination of hardware and software created by humans, is ultimately controllable by its human developers. He argued that the industry already possesses sufficient legal and regulatory frameworks to govern AI, negating the need for entirely new legislation.
Huang stressed that AI developers and companies building powerful AI models must prioritize the safety of their products before public release. “If we’re not confident about the safety of the products... then don’t release it,” he advised, extending this principle to all product-developing companies. He believes market forces inherently push companies to regulate their own development pace based on safety and readiness.
He further elaborated that companies should pause AI development if they deem a project uncontrollable or a product unsafe, resolve the issues, and then proceed. This perspective aligns with his view that existing market dynamics and corporate responsibility are more effective than external regulatory bodies in ensuring AI safety.
Meta CEO Mark Zuckerberg has echoed a similar sentiment in the broader AI slowdown discussion, highlighting the importance of “trust and alignment” in AI models. Zuckerberg suggested that labs failing to prioritize alignment and safeguards would ultimately fall behind, reinforcing the idea that internal commitment to safety is a critical differentiator.
Huang's remarks offer reassurance amid growing concerns about AI risks, positing that since humans build AI, they retain the ultimate control and responsibility for its safe deployment.