Amazon CEO flagged Anthropic AI risks to US officials

Amazon's CEO reportedly flagged cybersecurity risks with Anthropic's AI models to US government officials. What's interesting is that this happened before Washington restricted access to these models. According to a report, Amazon researchers used Anthropic's Claude Fable 5 model to get information that could be used in cyberattacks. Following these discussions, the US government imposed export-control restrictions on Anthropic’s Fable 5 and Mythos 5 models. An Amazon spokesperson confirmed that governments often seek their counsel on security risks, but they don't share details. The spokesperson also noted that Amazon Web Services has been affected by these restrictions. A former White House AI official claimed a trusted partner identified a vulnerability in the model. He alleged that Anthropic's CEO was asked to fix it or withdraw the model, but declined. However, Anthropic stated that the capabilities regulators cited are already available in other public AI models. The bottom line is that concerns about AI model security are leading to government action.