On June 12 (local time), the U.S. government, invoking national security considerations, issued an urgent directive to AI firm Anthropic. It demanded that Anthropic promptly revoke all foreign users' access—including non-U.S. employees located within the country—to its cutting-edge AI models, Mythos 5 and Fable 5. This move came a mere 72 hours after the models' release. Consequently, Anthropic had no choice but to disable these models worldwide, causing users to default to the older version, Claude Opus 4.8. In a statement, Anthropic contended that the so-called 'jailbreak vulnerability' cited by the U.S. government was merely a minor, industry-recognized flaw. The company argued that recalling deployed commercial models on such grounds was unwarranted. It also cautioned that if this precedent were set across the industry, it would impede the rollout of state-of-the-art AI models. On June 15, the European Commission announced that it was assessing the real-world implications of the U.S. export control directive. It underscored that emergency measures should not unfairly target partners and highlighted the incident as a wake-up call for Europe to bolster its technological self-reliance.