A warning from Amazon reportedly led the White House to shut down Anthropic's Mythos model
Fortune • • general
Key Points:
- Amazon researchers discovered a security bypass in Anthropic’s Mythos-class AI model, Fable 5, prompting concerns raised by AWS CEO Adam Jassy to senior U.S. officials and leading to a national security investigation.
- The U.S. Commerce Department imposed export controls barring Anthropic from distributing Fable 5 and Mythos 5 to foreign nationals, effectively disabling the models for all users due to the broad scope of the restrictions.
- This marks the first time the U.S. government has used export controls to halt access to a commercial AI model already in public use, sparking global debate over sovereign AI and national control of critical technology infrastructure.
- Anthropic and the U.S. government engaged in tense negotiations, with Anthropic disputing the severity of the security issue and the government citing refusal to remediate the jailbreak as justification for the export controls.
- The incident has prompted European politicians to call for increased AI sovereignty, highlighting risks of dependence on foreign AI systems and framing AI infrastructure as critical national infrastructure akin to electricity or the internet.
