At this week’s Salesforce Dreamforce conference, NVIDIA founder and CEO Jensen Huang Made a highly controversial claim: AI does not need new regulatory frameworks; safety issues should be handled by enterprises themselves. In his view, AI is not some “alien intelligence”—it is simply a human-built software and hardware system, fully controllable using existing legal and engineering tools.
“Safety is an engineering problem, not a legal one.”
Huang Renxun’s exact words were direct:“Safety is an engineering problem, not a legal one.” He repeatedly stressed that AI is, at its core, a complex computing system—since it consists of software and hardware, it can absolutely be designed, tested, and controlled by humans, without requiring dedicated legislation.
He also believes the free market itself provides sufficient constraints. “If you lack confidence in your product’s functionality, capabilities, or safety, don’t ship it—that’s self-evident.” In his vision, market pressure drives enterprises to self-regulate; innovation speed and product safety are not mutually exclusive—“running faster” and “building safer” can coexist.
Underlying stance: Interests and risks
Huang Renxun’s position is not without merit—among the world’s foremost experts on AI hardware, his understanding of “AI controllability” carries weight. Yet critics point out this stance is exceptionally “comfortable” for NVIDIA: the less regulation, the smoother the sales of AI hardware and software.
Real-world counterexamples abound: A CrowdStrike software update in 2024 grounded thousands of flights; Meta recently paid $18 billion to settle claims of “harm to children on its platform”; and the AI field has seen incidents including an OpenAI model breaching Hugging Face and users dying by suicide after prolonged conversations with chatbots. These cases demonstrate that even when companies act with “good intentions,” their products can still produce unintended consequences—relying solely on market self-regulation may not be reliable.
The industry’s window for self-regulation
Notably, Jensen Huang did not address another emerging path—Industry self-regulationThat same week, OpenAI, Anthropic, and Google were reported to have held weeks of private negotiations on AI safety and are considering “embedding” safety assessment teams within their respective organizations. This underscores that the AI industry itself has already recognized it cannot wait for legislation and must proactively establish some safety framework.
Jensen Huang’s stance may represent part of the industry’s interests, but his personal influence—such as his ability to “speak directly with President Trump”—is also significant. The AI regulation博弈 has only just entered deep waters.
Summary: The essence of the regulatory debate
Jensen Huang’s “no regulation needed” position is fundamentally a disagreement over “who bears responsibility for AI safety”: whether to trust corporate self-governance and market forces, or to introduce government legislation. For ordinary users, the answer may not be black-and-white—we neither want innovation slowed by overregulation nor safety vulnerabilities left unaddressed. The trajectory of this debate will directly shape the release cadence and trustworthiness of AI products over the coming years.
Follow AI industry developments and tool reviews—visit AI Dash—Discover the most useful AI tools.