From Silicon to Software: U.S. Extends Export Controls to AI Models

The sudden government seizure of Claude Fable 5 marks a watershed moment: AI export controls have shifted from controlling hardware to pulling deployed software, making model availability a geopolitical risk rather than a technical question. While this creates genuine precedent for how frontier AI will be governed, the security justification—a known jailbreak technique affecting vulnerabilities that plague competing models—appears inflated, masking regulatory theater with legitimate concern.