Amazon’s Jassy Raised Concerns About Anthropic Model Before Trump Crackdown
The Takeaway
- Amazon CEO Jassy among tech leaders to raise concerns over model.
- Trump administration restricted Anthropic’s advanced AI models for security.
- Restrictions followed a discovered “jailbreaking” method for Fable 5.
Amazon CEO Andy Jassy was among the tech leaders who raised concerns to senior Trump administration officials this week about security risks in Anthropic’s most advanced models, according to two people familiar with the conversations.
The calls between the head of Amazon—one of Anthropic’s biggest investors and vendors—and the officials in the last few days helped set in motion the Trump administration’s new export restrictions on Anthropic’s Claude Mythos 5 and Fable 5 models late Friday night, citing national security concerns, the people said. Those restrictions suspend access to those models to foreign nationals. Anthropic said it disabled access to the models for all customers to comply.
An Amazon spokesperson told The Information: “As a leading cloud provider that serves a large number of private and public sector customers, it’s not uncommon for governments to seek our counsel on potential security risks. When they occur, we don’t share the details of these discussions.”
The restrictions are the latest hit to Anthropic in its ongoing standoff with the Trump administration. The relationship between the Claude maker and the government has devolved in recent months following the Pentagon’s move in March to designate the company’s model as a supply chain risk. That move came after the two disagreed on whether Anthropic’s models could be used in cases like mass domestic surveillance or with lethal autonomous weapons.
The White House is unlikely to extend export restrictions on Anthropic’s advanced models to other AI companies, an official close to the U.S. government said Saturday.
Anthropic said Friday it had worked with U.S. and international government agencies to identify security vulnerabilities in Fable. The company had previously said it was concerned that the AI systems had become so advanced that they could hack software programs better than humans. The government did not object to the release of Fable, according to a source close to Anthropic.
Jassy’s conversations and the administration’s move to limit access to Fable 5 and Mythos 5 also highlight growing concerns from researchers and cybersecurity experts around the ability for Anthropic’s latest models to find new security vulnerabilities and exploits. Anthropic on Tuesday released Claude Fable 5, a neutered version of Mythos that’s blocked from being able to answer questions about cybersecurity and advanced AI research.
In a blog post published on Friday night, Anthropic said that it believed the government had “become aware of a method of bypassing, or ‘jailbreaking’ Fable 5” and its safety guardrails.
Anthropic cited a demonstration of how the jailbreak could be used to find vulnerabilities, but said that those vulnerabilities that were found were “relatively simple” and that other publicly-available models were able to find them as well.
After the discussions with Jassy and other tech leaders, U.S. national cyber director Sean Cairncross called a meeting on Friday with senior White House officials who decided that export restriction was the most straightforward way to take action against Anthropic, according to two of the people.
A spokesperson for the Office of the National Cyber Director did not immediately respond to a request for comment.
After the meeting, Anthropic CEO Dario Amodei and administration officials spoke on Friday, said four people familiar with the call. During the conversations, Anthropic officials laid out how the security vulnerabilities found through the alleged jailbreak were relatively simple and could be achieved with other models. But the government told Anthropic that it had already decided to implement the export control.
Anthropic and Amazon have a deep, multi-faceted relationship. Amazon has invested $13 billion in the company since 2023, and planned to invest another $20 billion in the future, tied to commercial milestones, the e-commerce giant said in April. Amazon’s Trainium chips help train and power Anthropic’s AI models, which also run on Amazon Web Services.
Amazon also has become a large investor in OpenAI this year, agreeing to invest up to $50 billion into the Anthropic rival and sell its latest models through AWS for the first time.
The Trump administration’s export controls follow an executive order it made two weeks ago to institute a voluntary framework in which AI companies could share their models with the government ahead of releasing them to partners and then the public.
The executive order was the result of a debate inside the White House on whether to put in place a more onerous licensing system for AI companies seeking to release advanced models, The Information previously reported.