Amazon CEO flagged Anthropic AI risks before US order

Amazon CEO Andy Jassy reportedly raised concerns about security risks in Anthropic's advanced AI models. These concerns were shared with senior administration officials. This happened before a government order forced Anthropic to take two of its newest models offline. Amazon has invested $13 billion in Anthropic since 2023. It plans to invest up to $20 billion more. Reports state Jassy told Treasury Secretary Scott Bessent that Amazon researchers used Anthropic's Fable 5 to get information that could help with cyberattacks. Amazon shared these findings with officials. An Amazon spokesperson confirmed that governments seek their counsel on potential security risks. However, they do not share details of these discussions. The government order cited a method to "jailbreak" Fable 5 to extract information for cyberattacks. This order suspended access for foreign nationals, forcing Anthropic to disable both Fable 5 and Mythos 5 for all users. Anthropic stated it disagreed that the situation warranted this action. The company believes the identified vulnerabilities were "relatively simple" and could be found using other public models. This situation highlights ongoing tensions between AI development and national security concerns.