On Saturday, Dario Amodei published his proposal to slow the pace of model development. On Monday, Donald Trump answered. On Truth Social, in capitals.
“The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” Then this: his administration has stopped “AI ‘people’ from doing bad, or potentially bad, ‘things,’ like Dario (Anthropic!), who is now pretending to be a ‘perfect little angel’.”
Hoax, scam, golden goose
One post wasn’t the end of it. Over the course of Monday, Trump wrote that the whole “AI/Data Center outburst” is happening only because the U.S. is leading. When, in the history of business, has anyone seen the leaders of an industry call for regulation that would drive them into bankruptcy? The concerns are a “hoax” and a “scam.” Plus: “Don’t kill the Golden Goose!”
As for the companies, he thinks Washington already holds the cards: “We already have tremendous CRIMINAL and REGULATORY power over these companies!” The post closes with a warning to “Conspiracy Theorists, Treasonists, Traitors, and Leakers.” Anthropic did not immediately respond to a request for comment from CNBC.
Sacks argues liability, not safety
The more interesting reasoning came from Trump’s AI czar. David Sacks had already posted on Saturday that the labs should go ahead and pace the frontier themselves – he just wants people to stop selling it as pure altruism.
His point: ship a model that enables a genuinely damaging cyberattack and you have a product liability problem. And the market already punishes models that behave unpredictably. A global agreement with China he considers “very unlikely.”
Nothing happens before the midterms
Practically speaking, the question is settled for now. The House leaves Washington on Thursday and stays in members’ districts through October, with the vote on November 3. Speaker Mike Johnson doubled down on CNN on Sunday: if Congress races into an emergency session to regulate AI, America loses the race to China. That fits what we saw in Washington last week.
Amodei picked the wrong opponent
There’s a gap in Amodei’s proposal, and Trump found it immediately. The plan relies on embedded evaluators, common standards and eventually international coordination – voluntary cooperation among rivals, plus government help for the rest. The White House just cancelled that help, and not on safety grounds. It cancelled it on competitive grounds.
What’s left of the proposal is the part Anthropic does alone: letting outside evaluators into the building. That isn’t nothing, but it isn’t what Amodei was after. His essay aimed at an industry-wide brake. What he got is a president calling him a little angel in public.
Sources: