Trump called the AI slowdown a hoax, live on Jensen Huang's speakerphone
Two days after Anthropic's CEO asked the industry to slow down, the President took a live call on stage at the All-In Summit and dismissed the whole idea. Here is what each response would actually change about how models ship, from the White House to the Speaker's office to China's foreign ministry.
A live call on stage, a duopoly argument from Sacks, a meeting from Johnson, a rebuttal from Beijing. 48 hours.
On Monday, September 14, at the All-In Summit in Los Angeles, Nvidia CEO Jensen Huang took a phone call from President Trump on stage and held it to the microphone. "The robots will not be taking over," Trump said, per Quartz's account of the exchange. "The AI will not be taking over the rest of the world. The whole thing is a hoax." Huang's reply, as TechCrunch reports it: "We're not going to let that happen, sir." The call was the loudest of four responses in 48 hours to Dario Amodei's essay asking the industry to slow down, covered here.
The White House: no slowdown
Trump framed the pacing proposal, TechCrunch reports, as something that would serve either political opponents or China. Huang agreed on stage that the two of them were "on the same page," Quartz reports, and said Nvidia would "make sure that everybody wins in the AI race in America." Quartz also records Trump saying data centers "make people wealthy" and calling AI "the oil of the next 20, 25 years." Effect on the proposal: the executive branch will not back the government-supported coordination the essay's second step needs.
Sacks: you are the frontier, so slow yourselves down
David Sacks, chair of the President's Council of Advisors on Science and Technology and previously the White House AI and crypto czar, posted his answer on X over the weekend. "Dario has written that we need to pace the frontier, and Sam has agreed," it begins. "People may be surprised by my response: go ahead. You guys are the frontier." By market share, revenue growth, and capability, he writes, OpenAI and Anthropic hold a duopoly on frontier intelligence, so if their unreleased models are too risky the two of them can slow down without anyone's permission. Benzinga quotes the follow-on: "The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability." Effect on the proposal: no antitrust cover, no waiver, and no regulator. Whatever pacing happens is voluntary.
The Speaker: a meeting, not a law
House Speaker Mike Johnson told Jake Tapper on CNN's State of the Union on September 13 that he would convene the leaders of roughly seven or eight AI platforms at the White House with lawmakers to set guardrails, and would do it "tomorrow" if he could, per Unite.AI's account of the interview. He rejected an emergency moratorium, said the executives now calling for guardrails are the ones pushing the frontier, and said safety work must not "smother American innovation." Axios summarizes his position as calling for solutions while saying Congress will not take the lead. Effect on the proposal: a meeting is on offer. A statute is not.
Beijing: fearmongering
China answered on the same Monday. Foreign ministry spokesperson Guo Jiakun, per The Decoder's report citing Bloomberg and the Financial Times, called for AI development that is "open, inclusive, universally beneficial and ethical" and said "fearmongering, confrontation and vicious competition" harm global governance. The state-run Global Times accused Amodei of waging a "silent AI Cold War." State Security Minister Chen Yixin, in the same report, warned of AI misuse by "hostile forces" through deepfakes and cyberattacks. Effect on the proposal: the essay's third step, coordination with authoritarian governments, has its first public answer, and it is no.
Why a build studio cares
We do not have a view on the politics, and this post is not the place for one. What we track is which of the essay's steps can still happen, because each one changes something concrete about the APIs our builds call. After this weekend, the count is one: the embedded evaluators, which need no government and which two labs have said they will host. The rate limit, the shared standards, and the international tiers all needed a counterparty in Washington or Beijing, and each counterparty said no within 48 hours.
For a client planning a product on a frontier API, that is the useful read: capability will keep arriving on the vendor's schedule, not a regulator's, and the only new thing to watch for is whether an outside evaluator's report on that vendor ever gets published. The moment one does, it goes into the vendor section of the audit.
Next step: read TechCrunch on the call, Sacks's post, Unite.AI on Johnson, and The Decoder on Beijing. If your roadmap assumes a model vendor's cadence and you want that assumption written down as a risk, write to us at hello@gattyworks.com.