Leadership2 publishers3 min readPublished
Huang calls the AI labs' demand for new rules a bid to escape the existing ones
Nvidia's Jensen Huang told CBS News there is a 0% chance AI ends the world in 2030, and that the lab chiefs asking Washington for guardrails want relief from the liability law already on the books. The White House agrees with him.
The Board Room · Leadership desk
What happened
- Nvidia's Jensen Huang told CBS News on Friday that 2030 will not be the end of the world and that there is a 0% chance of it, in an interview with correspondent Jo Ling Kent.
- He said the AI leaders calling for government regulation are not actually asking for more laws but asking to be relieved of the laws that already exist.
- Last week Anthropic's Dario Amodei published an essay titled "We Must Pace the Frontier" setting out a three-pronged plan to keep AI under the control of human engineers.
Compiled by The Board RoomSomething wrong?How this is made
Why it matters
- decision A company drafting its AI policy this quarter has to pick between the position of the firm that sells the compute and the position of the firms that buy it, and defend the choice to a board that will ask which one it followed.
- exposure Huang tied Nvidia's own valuation to safe deployment, so every customer incident involving models running on its hardware now tests a standard its chief executive set himself.
- precedent Self-halting has become the administration's first answer to any request for federal guardrails, and a lab that keeps shipping while warning of catastrophe will be asked why it did not take that option.
Huang's claim is a legal one: American laws on product liability and unauthorized entry already cover the harms in question [8]. That is the part of this a company can act on, and it is narrower than extinction. "We don't need more regulations," he said. "We need to apply the current regulations we have" [13]. He pointed to cyber exposure in particular: "You have all kinds of liabilities associated with cybersecurity" [9]. If that reading is right, the controls a buyer has before any new statute exists are the ones it can negotiate: indemnities, and the cyber cover it already pays for.
Leading AI chief executives said last week that development should be slowed until adequate risk-prevention measures are in place [14]. Anthropic's Dario Amodei set out three prongs in an essay titled "We Must Pace the Frontier," one of which asked the US government to regulate frontier labs "for antitrust reasons" [15][16]. Altman and Musk amplified it [17]. Huang described the request differently. "They're actually not asking for more laws. They're asking to be relieved of the laws we do have," he said [11]. He also said the lab chiefs "must be doing it for ulterior reasons," and that he did not know what those reasons are [12].
A skeptic's version of Huang's position is short: the company selling the hardware does not want a speed limit. CBS asked him why Americans should trust him on safety given Nvidia's incentives [7]. "Our company's success is directly connected to the safe deployment of products and services," he said. "If we don't continue to do that, our value would be diminished" [7]. That is a claim about aligned interest, not evidence about how the models behave. The same test cuts the other way, because federal supervision of frontier developers would bind the competitors of the labs requesting it [16].
The dividing line follows the supply chain. Nvidia's chips power the large language models that OpenAI and others build, and demand for them has made the company the world's most valuable at $5.3 trillion [6][5]. Nvidia, which sells the compute, says the existing rules suffice, and several of the buyers are asking to be constrained [1].
Federal policy currently sits on Huang's side. On Saturday, Trump wrote on Truth Social: "We will not in any way hinder or stifle the Growth of this incredible Industry" [18]. Michael Kratsios, who directs the White House Office of Science and Technology Policy, told Fox News on Sunday that the labs could "just stop" if they think their products are dangerous [19]. Huang repeated his argument that a slowdown among US labs would cede ground to China [21].
Huang said the predictions are "not grounded in science" [4], and the document on the other side is a proposal for safeguards [15]. The two accounts differ on one detail: Business Insider identifies Amodei as Anthropic's chief executive [15], CBS as the company's former chief executive [24]. A policy paper that cites the industry position now has to say which chief executive it means.
What to watch
- Whether Kratsios's "just stop" answer becomes a written federal position or stays an interview line.
- Whether any frontier lab pairs its slowdown call with commitments a customer could enforce in a contract.
- What comes of Huang's stated intent to discuss global AI standards with Xi Jinping at the Sept. 24 White House dinner.