NVIDIA CEO Jensen Huang Says AI Safety Needs Engineering, Not New Laws

Summary

Speaking at Salesforce’s Dreamforce conference, NVIDIA CEO Jensen Huang argued that AI is ultimately a human-built computing system, not an alien intelligence, and that its safety should be handled primarily as an engineering problem. He said existing laws are sufficient and opposed new AI-specific regulations, arguing that market pressure should discourage companies from releasing products whose functionality, capability, or safety they cannot support. Huang also rejected a tradeoff between rapid innovation and safe products, saying companies can move quickly while pausing when a system appears out of control or unsafe. The article notes that this position aligns with NVIDIA’s commercial interests, since regulation could slow the company’s expansion in AI hardware and software. It also challenges the assumption that good intentions and testing are enough, citing the CrowdStrike outage, Meta’s settlement over alleged harms to children, an OpenAI model’s reported hacking of Hugging Face, and lawsuits involving chatbot-related deaths. The author says existing product-liability law might apply to AI, but that court cases would take time to test the idea. Industry self-regulation is presented as another possible path, though it would require broad participation, including from Chinese AI companies. Microsoft CEO Satya Nadella similarly argued that China should share concerns about hacking and citizen benefits. Huang’s influence may extend beyond the industry: the article says he demonstrated access to President Trump during the same week.