Amazon CEO Andy Jassy may have contributed to security concerns that led Anthropic to restrict access to two of its models, according to reports. The Wall Street Journal cited internal discussions where Jassy informed Treasury Secretary Scott Bessent and other officials that Amazon researchers had used Anthropic’s Claude Fable 5 model to access data that could be exploited for cyberattacks. The government subsequently imposed an export control ban on the Fable 5 and Mythos 5 models. An Amazon spokesperson stated that while the company regularly consults with governments on potential security risks, it does not disclose the specifics of these discussions.
The spokesperson also noted that AWS has been impacted by the model restrictions. The Information and Reuters reported that Amazon, as a major investor in Anthropic, had raised concerns about the security of its models. David Sacks, former Trump AI czar and current member of the President’s Council of Advisors on Science and Technology, provided an account of the discussions, claiming that a trusted partner of both Anthropic and the U.S. government reported a 'jailbreak' in the models. Sacks added that the administration requested Anthropic CEO Dario Amodei to address the issue or de-deploy the model, which Amodei refused to do.
Anthropic later stated in a blog post that the features causing government concern are already available in other publicly accessible models. Source: techcrunch
