Nvidia founder and chief executive Jensen Huang reckons artificial intelligence needs better engineering, not another fresh pile of government regulation.
According to TechCrunch speaking at Salesforce’s Dreamforce conference in San Francisco, Huang argued that AI remains software running on human-built computing systems, not some mysterious intelligence beyond human control.
“Safety is an engineering problem, not a legal one… developers can control AI systems using engineering practices, existing laws and the same product disciplines applied elsewhere in technology,” he said.
Huang said companies should simply avoid releasing products when they are not confident those products work properly or can be operated safely.
He argued that market pressure already gives companies an incentive to avoid shipping dangerous products because customers are unlikely to remain enthusiastic when expensive software starts causing expensive disasters.
“We don’t need any new laws. We don’t need new regulations… innovation speed and product safety do not have to be competing objectives.”
His position puts Nvidia on a different side of the current industry debate from Anthropic chief executive Dario Amodei. Amodei has called for stronger safeguards and coordination around increasingly capable frontier models.
OpenAI, Anthropic and Google DeepMind have meanwhile been discussing ways to cooperate on AI safety, according to reports Reuters cited. OpenAI has expressed support for bipartisan legislation to address catastrophic AI risks.
Huang’s argument rests partly on the idea that existing product liability and other laws remain applicable when AI causes harm. Critics of relying on existing rules argue that increasingly autonomous models could create risks that current legislation was never designed to handle.
A fairly obvious commercial dimension lurks behind the philosophical discussion. Nvidia supplies much of the hardware powering the AI boom, making rapid expansion of AI infrastructure extremely good news for the outfit selling the expensive silicon underneath it.
Huang remains bullish about how far that expansion can run. At Dreamforce, he said his ambitions for Nvidia had increased as AI improved productivity, while arguing that the same technology could expand opportunities across industries and countries.
The debate is becoming increasingly divided among technology executives. Huang and Meta chief executive Mark Zuckerberg have resisted coordinated efforts to slow AI development, while Amodei, OpenAI chief executive Sam Altman and others have argued for stronger safety measures as model capabilities increase.
