Anthropic CEO Dario Amodei famously warned that his products were a threat.
In an essay published this month, Amodei warned that the power of AI “has become undeniable.” He cited Anthropic’s latest model, the Mythos, as an example, saying it poses a “very real risk” to cybersecurity, the financial sector, critical infrastructure and national security.
He also called for stronger government intervention to address the risks.
His warning appears to have backfired Friday, with the U.S. government’s sudden intervention.
Anthropic announced that the Trump administration has ordered it to cut off access to the models from foreign countries, cutting off access to its Mythos 5 and Fable 5. The Pentagon’s chief information officer expressed support for the move in an X post, writing, “Some things simply matter more than revenue cycles, clickbait, and pre-IPO valuations.”
The surprising development caused an uproar in the technology industry.
Gary Marcus, an AI researcher and notorious skeptic, said the Trump administration’s move is “hugely excessive and counterproductive.”
Yann LeCun, considered one of the godfathers of AI, criticized Amodei, writing in X that “Dario Amodei’s ridiculous fear-mongering about myths and fables (and AI in general) has finally paid off.” “A man reaps what he sows.”
Amodei has long sought to position himself as an adult in the AI industry. Once a key researcher at OpenAI, he left the company to found Anthropic over concerns that OpenAI CEO Sam Altman was prioritizing product releases over ensuring product safety.
Since then, he has made a series of statements along with OpenAI, Google, Meta, Amazon, and others saying that what he is building and making available to the public could forever change our lives, for better or worse.
Amodei previously warned that AI would eliminate half of entry-level white-collar jobs and push unemployment to levels not seen since the COVID-19 pandemic and before that, the Great Recession of 2008.
Although Mr. Amodei has softened his tone on the job in recent weeks as the company prepares to go public, he and his company have not held back on other safety concerns.
“The cyber risks posed by the Mythos class of models are not the last risks we will have to face,” he wrote in a June essay. “I believe biological risks may soon follow, and serious AI autonomy risks may occur before too long.”
He wrote that the lawmakers’ concerns are out of proportion to rapid advances in AI.
Earlier this month, his company called for a moratorium on development of frontier AI models in a paper warning that the latest models were approaching self-improvement and “could increase the risk that humans will lose control of AI systems.”
“We believe it’s good for the world. option “Delay or temporarily halt cutting-edge AI development so that social structure and coordination research can catch up with technological advances,” Antropic wrote.
That pause, at least for Anthropic, came on Friday.