Jensen Huang, Nvidia’s founder and CEO, expressed a clear position on AI regulation during his appearance at Salesforce’s Dreamforce conference: In his view, there is no need for new AI-specific laws or regulations because the safety of these systems can be treated as an engineering problem.
Huang said AI is not a “strange mind” separate from human understanding, but ultimately a combination of hardware and software developed by humans. Accordingly, he believes developers can control it using engineering methods and existing laws. He added that companies should not release a product unless they are confident in its functions, capabilities, and safety.
The Market Argument Instead of New Rules
Huang believes market forces are sufficient to pressure companies to proceed cautiously before releasing unsafe products. Under this view, companies can continue innovating quickly, but they are expected to pause when they feel that the product or organization has gone out of control. He also rejected presenting speed, innovation, and safety as mutually exclusive options, emphasizing that they can all be achieved together.
This position defends companies’ ability to self-regulate their products and aligns with Huang’s call to use open-weight models as a competitive counterbalance to laboratories developing proprietary models. However, the article points to another path that Huang did not discuss in detail: industry self-regulation, with a limited window for AI laboratories around the world, including Chinese laboratories, to participate in shared safety rules.
What Does This Position Raise?
TechCrunch’s analysis acknowledges that treating safety as an engineering issue may seem logical, especially since AI systems are software and computing products. But it argues that engineering tests and market forces do not always prevent defective products from being released or unintended consequences from occurring. In this context, it cites the 2024 CrowdStrike outage, which caused blue screens, disrupted thousands of flights, and interfered with other businesses.
The article also states that companies may act, according to the allegations it cites, in ways that do not always align with the public interest. It points to Meta’s payment of $18 billion to settle a lawsuit related to social media’s harms to children, as well as harms attributed to AI systems, including the breach of an OpenAI model connected to the Hugging Face platform and lawsuits against the laboratory over the suicides of people who had lengthy conversations with its chatbot.
Between Legal Liability and Self-Regulation
The article raises the possibility that existing product-liability laws may cover some AI-related harms, but considers that testing this possibility in practice could take a long time through the courts. This leads to an open question: Is it enough for companies to be held responsible for their products after harm occurs, or does the industry need shared preventive rules before launch?
The discussion also has an international dimension. According to the article, Microsoft CEO Satya Nadella said at the All-In Summit that China should care about the same safety risks as the United States, such as hacking and protecting citizens. At the same time, the source links Huang’s position to Nvidia’s commercial interests and its ambition to continue selling AI systems and software, noting that he said the ceiling for ambition and growth associated with the productivity AI provides is extremely high.
certi.news analysis: The real change here is not the launch of a new technology, but an attempt to influence the shape of future AI governance. Huang’s position places the burden on engineering teams, corporate decisions, and market forces, while the examples cited by the source serve as a reminder that good intentions and testing do not necessarily eliminate failures or harms. The article does not determine whether current laws are sufficient, nor does it offer a practical model for self-regulation; therefore, the limits of liability and mechanisms for international cooperation remain among the most prominent open questions.