Anthropic has abruptly disabled its most advanced artificial intelligence models following a U.S. government order restricting foreign access due to national security concerns. The company confirmed it must suspend Fable 5 and Mythos 5 for all users to comply with the Commerce Department directive.
Officials cited fears that foreign nationals could bypass safeguards to identify software vulnerabilities. Anthropic disputes this assessment, stating the government provided only verbal evidence of a narrow jailbreak method. The company argues that recalling a commercial model deployed to hundreds of millions based on unverified claims sets a dangerous precedent for the industry.
This action marks a significant escalation in U.S. export controls, shifting focus from hardware restrictions to limiting access to AI capabilities themselves. Experts warn these Mythos-class models could accelerate sophisticated cyberattacks against critical banking infrastructure if misused.
The directive complicates Anthropic’s relationship with federal regulators as the company prepares for an IPO. Amazon Web Services confirmed it is revoking access across all regions per Anthropic's request. Industry observers note this standard could effectively halt new frontier model deployments if applied universally.