Anthropic CEO Dario Amodei is famous for warning that his own products pose a threat.
In an essay published this month, Amodei cautioned that AI’s power “has become undeniable.” As an example, he pointed to Anthropic’s latest model, Mythos, which he said presents “very real risks” to cybersecurity, the financial sector, critical infrastructure, and national security.
He then called for a more robust government intervention to address the risks.
On Friday, his warnings appeared to backfire when the US government suddenly intervened.
Anthropic cut off access to its Mythos 5 and Fable 5 after it said the Trump administration ordered it to block foreign access to the models. The Pentagon’s chief information officer expressed support for the move in an X post, writing, “Some things are simply more important than revenue cycles, clickbait, and pre-IPO valuation.”
The surprise development sent the tech industry into a tizzy.
Gary Marcus, the AI researcher and notorious skeptic, called the Trump administration’s move “wildly overdramatic and also counterproductive.”
Yann LeCunn, who is considered one of the godfathers of AI, blamed Amodei: “Dario Amodei’s ridiculous fear mongering about Mythos/Fable (and AI in general) finally pays off,” he wrote on X. “One reaps what one sows.”
Amodei has long sought to position himself as the AI industry’s adult in the room. Once an integral researcher at OpenAI, he left the company to start Anthropic over concerns that OpenAI CEO Sam Altman prioritized releasing products over ensuring their safety.
He has since issued a stream of statements that the things he is building and releasing to the general public, alongside OpenAI, Google, Meta, Amazon, and others, could upend life as we know it — for good as well as bad.
Amodei once warned that AI would eliminate half of entry-level white-collar jobs and cause unemployment to soar to levels not seen since the COVID-19 pandemic, and before that, the Great Recession in 2008.
While Amodei has softened his tone on jobs in recent weeks as his company gears up to go public, he and his company have not held back on other safety concerns.
“The cyber risks that Mythos-class models present will not be the last that we must face,” he wrote in his June essay. “I believe that biological risks may soon follow, and that serious AI autonomy risks may not be far behind.”
He wrote that concerns by lawmakers were out of step with AI’s rapid progress.
Earlier this month, his company called for a temporary halt to the development of frontier AI models in a paper cautioning that the latest models are getting closer to improving themselves, which it said, “might increase the risks of humans losing control over AI systems.”
“We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology,” Anthropic wrote.
That pause arrived on Friday, at least for Anthropic.
