- Published on
We don’t need AI regulation — leave safety to us, Nvidia’s Jensen Huang says
- Authors
- Name
- DeepHorus
Nvidia founder and CEO Jensen Huang used a keynote stage at Salesforce's Dreamforce conference on Tuesday to argue that artificial intelligence does not require new regulation. His comments, made during a wide-ranging interview, amount to a direct challenge to lawmakers and safety advocates pushing for binding rules on frontier AI. Huang's position matters because Nvidia is the dominant supplier of the GPUs that power modern AI, giving him unmatched influence over the industry's direction.
The debate over AI safety has intensified since ChatGPT's launch. Policymakers in the US and EU have proposed risk-based frameworks, transparency requirements, and licensing for advanced models. Some researchers warn about existential risks, with one OpenAI safety researcher describing AI as an "alien mind." Huang rejects that framing, saying AI is built by humans and remains a computing system.
At Dreamforce, Huang said safety is an engineering problem, not a legal one. "We're developing software after all," he said. "It's a complicated computing system, but it's ultimately a computing system." He argued that market forces already pressure companies to avoid unsafe releases. If firms lack confidence in a product's safety, they should simply not ship it, he said, pacing themselves until the market appreciates the result. Nvidia itself sells chips, systems, open-source models, agents, harnesses, and sandboxes, making it central to AI deployment.
Huang's argument rests on a narrow technical view: AI models are software artifacts trained on data and run on hardware. Safety, in this view, can be addressed through testing, red-teaming, staged deployment, and engineering controls. But critics note that large models can behave unpredictably, generate harmful content, or be repurposed for cyberattacks and disinformation. Existing laws cover some harms, yet they may not address emergent capabilities or rapid release cycles.
In the short term, Huang's stance could embolden AI companies to resist new compliance burdens, especially in the US where federal rules remain fragmented. In the long term, it may widen the gap between voluntary safety practices and calls for enforceable standards. Nvidia benefits from faster AI adoption, so regulation that slows deployment could reduce demand for its chips. That commercial incentive makes his deregulatory message unsurprising to skeptics.
Industry reaction was mixed. Some executives welcomed his optimism and emphasis on innovation speed. Safety advocates and some researchers pushed back, arguing voluntary guardrails have repeatedly failed in social media and other tech sectors. Market analysts noted Nvidia's stock has soared during the AI boom, giving Huang enormous credibility but also a conflict of interest. Even sympathetic observers questioned whether "run as fast as you can" is compatible with responsible deployment.
Huang said his ambition is greater than ever, calling the sky the limit for Nvidia, every industry, and every country. Regulators, however, are unlikely to abandon oversight. Expect continued battles over model evaluations, transparency, and liability. The key question is whether engineering culture and market pressure alone can prevent serious harms—or whether a high-profile AI incident will force the regulation Huang insists is unnecessary.
