The government used a private "is informed" letter to apply deemed export controls, which regulate a foreign national's access to technology *within* the US. This powerful tool effectively halted the Fable model's use, even by Anthropic's own foreign national employees, without public rule-making or debate.

Simple refusal mechanisms in AI models are easily bypassed by motivated actors. Effective biosecurity requires deeper interventions, such as curating training data to exclude sensitive biological information or implementing strict access controls for the most powerful models, ensuring they aren't publicly available.

The recent restrictions on allies signal a shift toward a tiered system for frontier AI models, similar to how advanced weaponry is shared. Top US government entities and companies will get first access, followed by a lower tier of close allies, who should not expect unfettered access to the latest American AI capabilities.

By unilaterally revoking access for all non-US nationals, the US government demonstrated that reliance on American frontier models is a strategic vulnerability. This single action validates the need for "Sovereign AI," powerfully motivating other nations to invest heavily in their own domestic AI capabilities to ensure technological independence.

Because AI models can be easily downloaded, traditional regulation is ineffective. The logical endpoint isn't policy, but active 'algorithmic warfare' where proprietary models are used to launch offensive attacks to degrade or trick competing open-source and foreign state-sponsored models.

The current US strategy is contradictory. While taking extreme measures to block allies like Canada from accessing advanced US AI models, the administration's inaction has left open loopholes that allow Chinese firms to freely acquire the very chips needed to build competing models. This highlights a critical disconnect.

U.S. AI strategy is incoherent. While the Treasury Department tightly controls domestic access to advanced models like Anthropic's Mythos for national security, the administration also facilitates Nvidia's sale of the very AI chips to China that will accelerate their ability to develop competing models.

The US strategy for controlling AI chip exports has evolved from blocking product sales to supervising entire networks. Authorities now focus on loopholes like foreign subsidiaries, third-country routing, and cloud access, signaling a more sophisticated approach to compute governance.

Sebastian Malabai argues that U.S. chip export bans are ineffective because China circumvents them by renting GPU capacity in other countries and using "distillation" to reverse-engineer and copycat advanced U.S. models. This suggests a need for a new strategy focused on collaborative safety.