Amazon’s CEO reportedly caused a government crackdown that shut down Anthropic’s super-powerful AI
The TL;DR
Amazon’s Jassy is reported to have told the government that researchers are using Fable 5 to get cyberattack information. That prompted the order to shut down Fable 5 and Mythos 5.
Amazon CEO Andy Jassy was reportedly the source of security concerns that led the US government to force Anthropic to shut down Fable 5 and Mythos 5 to all customers on Friday. The Wall Street Journal reported that Jassy told Treasury Secretary Scott Bessent and other officials that Amazon researchers were using Fable 5 to obtain information that could be used in cyber attacks. The government then imposed a ban on export controls for both types.
The revelation adds an uncomfortable element to the story. Amazon is one of Anthropic’s biggest investors, having invested billions and secured a $100 billion cloud investment commitment. The infrastructure company Anthropic is the one that has told the government that its species are dangerous.
An Amazon spokesperson said “it is not uncommon for governments to seek our advice on potential security risks” but declined to share details of the negotiations.” The spokesperson also pointed to an AWS status update confirming that Amazon’s cloud platform is affected by the model shutdown.
David Sacks, Trump’s former AI chief who now co-chairs the President’s Council of Advisors on Science and Technology, offered a different account. He said “A very reliable partner of Anthropic and USG came forward with a jailbreak.” Sacks added: “The administrator asked Dario to fix the jailbreak or stop using the model. Dario refused.“
The Anthropic version is different. The company said it reviewed the jailbreak process and found that it had already “a small number of previously known vulnerabilities, which are small.” It called the government’s response disproportionate and said that the power that caused concern was already found in some publicly accessible models. The shutdown affected all customers worldwide because Anthropic cannot filter foreigners from US users in real time.
Consistency is important in the AI industry. Amazon has invested billions in Anthropic. Models built with Anthropic on AWS. Amazon’s CEO told the government that those brands pose a security risk. The government used export controls to force them off the Internet. AWS was then affected by the shutdown. It’s the ouroboros of corporate dependency and regulatory blowback.
For Anthropic, the immediate question is how quickly it can return access. In the wider industry, example is key. If a major cloud provider could trigger export control action against its portfolio company by raising concerns with the Treasury Secretary, the competitive forces in the AI market just got a new weapon. Mythos is used by banks and government agencies to detect risk. All of those customers lost access because of the dispute between Anthropic’s biggest investor and the government whose CEO has been seeking more regulatory powers.
