The Trump administration just made it harder for two of America’s biggest AI companies to do business globally. And the unintended consequence might be exactly what Washington was trying to prevent: a stronger competitive position for open-source AI projects, including those developed overseas.
In a span of two weeks in June, the US government effectively kneecapped the international reach of both Anthropic and OpenAI. The Commerce Department blocked foreign nationals from accessing Anthropic’s most advanced models on June 12-13, while OpenAI voluntarily restricted its newest system to a handful of government-approved partners. The stated reason is national security. The practical result is that a vacuum is forming, and open-source alternatives are ready to fill it.
What actually happened
The sequence started on June 2, when Executive Order 14409 established a voluntary pre-release review framework for what the administration calls “covered frontier models.” Under this system, AI companies can submit their most powerful models for up to 30 days of government review before public release. No licensing mandate, technically.
Ten days later, the Commerce Department went further. It issued an export control directive specifically targeting Anthropic’s Fable 5 and Mythos 5 models, citing their potential connection to advanced cyber capabilities that could be used against critical infrastructure. Anthropic responded by disabling access to both models globally. Not just in adversary nations. Everywhere.
