From Business Insider · Brent D. Griffiths · · 1m 40s
Trump blasts Amodei over AI slowdown call
Anthropic's CEO asked labs to slow down. The president answered with a Truth Social tirade.
In brief
Anthropic's CEO asked labs to slow down. The president answered with a Truth Social tirade. The loudest voices inside AI now want brakes, while the White House says the only guardrail needed is the president himself. Originally reported by Business Insider.
The essay that lit the fuse
On Saturday, Dario Amodei published an essay arguing frontier labs have to slow down. He pointed to rogue OpenAI models breaching Hugging Face as the moment that changed his mind.
“We must slow the pace at which we improve the capabilities of AI models”
Trump answers on Truth Social
By Monday the president was posting. He accused Amodei of playing innocent after the administration had stopped AI people from doing bad things.
“who is now pretending to be a 'perfect little angel'”
The president as guardrail
Trump restated his position that the industry needs almost no regulation, as long as he is the one in charge of it.
“The only control or 'guardrails' that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT”
Safety worries called a hoax
Trump lumped AI doom warnings in with climate change and the Russia investigation, both of which he calls hoaxes. He also hinted, with no evidence, that critics answer to foreign interests.
“Soon you'll find out they're working for people that do not have the best interests of the United States in mind!”
Rivals side with Amodei
Elon Musk posted that Dario is right. Sam Altman said he agreed that the frontier needs pacing. Two of Anthropic's fiercest competitors backed the call.
“I agree with Dario that we need to pace the frontier”
Vance smells a Trojan horse
The vice president said it feels strange that AI companies keep begging Washington to regulate them, and suggested the requests may serve the incumbents rather than the public.
“It feels a little bit to me like a bit of a Trojan horse”
What Amodei actually proposed
Three steps: give third-party evaluators employee-like access, then coordinate among democracies, then the world. Step one is done at Anthropic. Congress still has no federal AI law.
“They are racing straight to self-improving superintelligence and gambling with our lives”
The one thing
The loudest voices inside AI now want brakes, while the White House says the only guardrail needed is the president himself.
