On June 12, the US Commerce Department ordered Anthropic to cut off its two newest models to every foreign national, leading to a worldwide shutdown of Claude Fable 5 and Mythos 5 within hours of a letter arriving at 5.21 pm. The stated concern was that researchers had found a way to prompt Fable 5 into identifying software flaws and writing code showing how one could be exploited. The controls were lifted on June 30. The article argues that a cyber-specific worry was answered by pulling a general-purpose product used by millions for harmless tasks, illustrating what happens when the only lever points at the model rather than the harm. In Delhi in February, Sam Altman told the AI Impact Summit that the world “may need something like an IAEA” to coordinate AI, later proposing a US-led standards forum in the Financial Times. Dario Amodei has argued for governance closer to aviation regulation. The UN convened its first Global Dialogue on AI Governance in Geneva on July 6-7, where an expert panel co-chaired by Yoshua Bengio warned that technology is outpacing governments’ ability to adapt. Beijing is assembling a World AI Cooperation Organisation, and India’s New Delhi Declaration, endorsed by 89 countries, runs to seven pillars with little on hard security. The article concludes that nuclear-style export controls are wrong because AI is a general-purpose technology with no chokepoints, and a single overarching regime is problematic.