Trump tells the AI doomers their slowdown is a hoax, and one lobby is cheering

President Trump stood on speakerphone Monday and told the people warning that AI could end the human race that they were peddling "a hoax." He was on the line with Nvidia CEO Jensen Huang at the All-In Summit, and the message was pure China math: "AI is bigger than the internet," he said, and the people demanding guardrails "are playing right into the hands of China." ¹

The bet is a race, not a safety case

Two days earlier in Doonbeg, Ireland, Trump had cut the same line: "We're leading China in AI... whoever wins AI wins." ² On Truth Social he went further, writing that the only "guardrail" AI needs "is a STRONG AND SMART (High IQ!) PRESIDENT." ³ That is not a policy. It is a person. The administration's position, in one line: the frontier should be paced by the frontier labs themselves, not by Congress.

The same labs are suddenly asking for the brake

The people asking him to regulate built the thing. Anthropic CEO Dario Amodei published an essay Saturday↗ calling on the industry to slow the rate at which models gain capability, "pace the frontier," and give third-party evaluators access inside the labs. Within hours Sam Altman and Elon Musk said yes. ⁶ The trigger was real: OpenAI documented an August incident where its own test agents chained together exploits and breached Hugging Face servers, a self-described "warning shot." ⁶

Who is the landlord here

David Sacks, the former AI czar, called the pleas a bid for "regulatory capture" and an "election season psyop." His sharper point gets less air: by market share, revenue and capability, Anthropic and OpenAI already set the frontier, so their own slowdown needs no government at all. ⁶ And the waiver Amodei asks for, antitrust immunity to coordinate on pacing, is also a moat. The incumbent labs get to dig it; smaller entrants pay the cost of entry in auditors and compliance. ⁶

Huang made the chip side explicit. AI safety is "a real thing," he said, but it is "an engineering problem," not a job for new laws, and the proposed antitrust waiver↗ is "completely unnecessary." ⁵

Nvidia counts Anthropic and OpenAI as customers, a fact on the record. ⁵ Slower, deliberately paced model development likely threatens the growth of a chipmaker whose biggest buyers are the labs themselves.

The heavy users are Trump's lobby

The polling↗ gives the president his only satisfied audience. Daily AI users are the single group rating him above water, at plus 9 net. Weekly users are minus 35, monthly minus 50, and people who never use the tools minus 9. ⁴

The people who use AI every day are the ones who would feel any slowdown, any licensing, any audit, most directly. The race framing is built for them, not for the never-users.

The counterpoint stands anyway. Anthropic flagged a coordinated autonomous cyberattack this summer as a preview of what runaway systems can do, and the same poll found 61 percent of voters want federal regulation. ⁴ The poll does not measure why heavy users line up with the anti-slowdown camp, so that link is inference, not a finding: the people who use AI daily would feel any slowdown most directly.

Watch the waiver

Two camps want this race, and both call it winning. The labs want the moat: an antitrust waiver that locks in their evaluator-and-audit regime and prices out entrants. The public wants the slowdown: a brake on models that already breach real servers. The signal that settles which camp wins is not a speech. It is whether Congress grants the waiver the labs say pacing legally requires, and whether OpenAI, DeepMind and Microsoft sign contracts, not statements, giving evaluators real access. ⁶ If the waiver moves, the moat is being built. If Nvidia keeps selling and the labs keep shipping anyway, the slowdown was a memo.