Amazon CEO reportedly raised Anthropic Fable concerns before US forces offline models – GeekWire
Amazon CEO Andy Jassy was reportedly among the tech leaders who contacted top Trump administration officials about security risks in Anthropic’s most advanced AI models, before a government order forced the AI lab to take its two new models offline.
The situation puts Amazon in an unusual and potentially difficult position with Anthropic, where it has invested $13 billion through 2023, with plans to invest about $20 billion more.
The Information first reported the phone calls between Jassy and senior officials, citing two people familiar with the discussions. The Wall Street Journal reported that Jassy told Treasury Secretary Scott Bessent and others that Amazon researchers used Anthropic’s Fable 5 to obtain information that could be used in cyberattacks.
Amazon shared the findings with administration officials, according to reports.
“As a leading cloud provider serving a wide range of private and public sector customers, it is not uncommon for governments to seek our advice on potential security risks,” an Amazon spokesperson said in a statement sent to GeekWire on Monday morning. However, the statement added, the company does not share details of these discussions if they occur.
The executive order, issued on Friday afternoon, revealed how to jailbreak Anthropic’s Fable 5 – a commonly used version of its more powerful Mythos 5 model – to extract information that could aid in cyber attacks. The order stopped access to any outsider, forcing Anthropic to disable both models for all users to comply.
Axios reported that Amazon was among at least five companies that raised concerns with administration officials Thursday night and Friday before the order came down.
In a statement on Friday evening, Anthropic said it was complying with the government’s order but did not agree that the situation warranted action. The company said the vulnerabilities identified using Fable were “simple” and found using other publicly available models, including OpenAI’s GPT-5.5.
“If this standard is implemented across the industry, we believe it will stop all new model shipments from all cross-border model suppliers,” the company said.
Independent experts questioned the seriousness of the result. Andrew Morris, founder of cybersecurity firm GreyNoise Intelligence, told the Journal that Amazon’s report showed that Fable could reveal security bugs in at least four software programs, but that the information is “far from dangerous cybersecurity information.”
Legend 5 remains unavailable to Anthropic’s Claude users as of press time.
It’s the latest twist in the contentious relationship between Anthropic and the Trump administration. Earlier this year, the Pentagon designated the company’s model as a supply-chain risk, after the two sides argued over whether Anthropic’s models could be used for purposes such as mass domestic surveillance or lethal autonomous weapons.
