Anthropic has pulled the plug on access to its latest flagship model, Fable Five, for non-US users—even its own international employees. The move came after researchers discovered the model could be jailbroken to reveal information useful for cyberattacks. While framed as a security measure, the restriction is sparking a broader reckoning: companies now realize they can't rely on a single country's AI infrastructure. Analysts predict this will trigger a wave of investment in 'intelligence sovereignty'—homegrown AI alternatives and self-hosted models that let countries and enterprises control their own intelligence layer rather than depend on American companies. It's a shift comparable to China's multi-year push to migrate away from U.S. technology—and a potential headwind for U.S. AI labs betting on global dominance.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton