Amazon raised worries about Anthropic AI models before US crackdown, source says

Amazon CEO Andy Jassy was among tech titans who raised security concerns to senior Trump administration officials this week over Anthropic’s most advanced AI models, a person familiar with the subject told Reuters.

Jassy’s presence underscores the dramatic step taken by Anthropic on Friday to shut down its newest models worldwide under national security demands from the administration of President Donald Trump.

The San Francisco-based AI startup, which has confidentially filed for a U.S. initial public offering, had previously warned about hacking capabilities of its Mythos model and held it back from wide release, but earlier this week, Anthropic rolled out a public version, called Fable, with what it described as cybersecurity safeguards.

That brief release was over Friday. In a blog post, Anthropic said the U.S. government notified the company it believes there is a way to bypass, or “jailbreak,” a protection ​against using the model to uncover cybersecurity flaws.

In a blog post, Anthropic said the bypass only revealed “minor” security weaknesses that other publicly available models may find.

The Trump administration told the business to prevent any foreign nationals, inside or outside the U.S., from utilising both its latest models, Fable 5 and Mythos 5, Anthropic stated. Anthropic responded by saying it would block access to the models worldwide.

Amazon will not say if it had spoken to government officials regarding Anthropic’s models.

“It is not unusual for governments to ask us for advice about potential security issues because we are a leading cloud provider serving a large number of customers in the public and private sectors,” an Amazon spokeswoman said. “When they happen, we don’t disclose the details of these discussions.”

EXPORT RESTRICTIONS
Earlier Saturday, tech news site The Information highlighted Jassy’s concerns. The Information later reported, citing a U.S. official, that the administration was unlikely to require other AI companies to adhere to limits comparable to those placed on Anthropic.

Reuters could not immediately confirm plans by the Trump administration to regulate other corporations.

The U.S. government’s prohibitions were an export control, Anthropic stated in its blog post. The Bureau of Industry and Security of the U.S. Commerce Department, which supervises export controls, did not react right away to a request for comment.

Officials issued the export control “reluctantly” after Anthropic CEO Dario Amodei “refused” to “fix the jail break or de-deploy the model”, White House adviser David Sacks stated in a social media post on Saturday.

“The hope now is ​that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general ​release,” wrote Sacks, ⁠co-chair of Trump’s Council of Advisors on Science and Technology and once the White House’s AI czar.

Some advocates of export limits were puzzled by the Trump administration’s move because it also applies to allied nations, not only rivals.

“This wasn’t thought out very well,” said Jimmy Goodrich, a senior scholar at the University of California’s Institute for Global Conflict and Cooperation. “It even prohibits Canadians and ​Brits working at Anthropic from doing research and development.”

The directive came as a previous fight between Trump administration officials and Anthropic was simmering down among portions of the U.S. ​government.