Is AI access being restricted by who you are?
Yes - but narrower and sharper than the fear. One real, unprecedented case did gate frontier AI by nationality. A country-rationing rule was proposed and killed. And 'ID to use AI' is a real direction, not a present law. Here is the whole map, and the way out.
In June 2026 the US ordered two frontier AI models suspended for EVERY foreign national worldwide - allies treated the same as adversaries. The provider took both models fully offline.
It caught a Mexican, a Canadian, a German the same way: as 'not American.' Citizenship-blunt - not aimed at any one country. The legal basis (the decades-old 'deemed export' rule) is real.
A rule sorting whole countries into tiers for AI-chip and closed-model access was issued in January 2025. Most of the world sat in the metered middle tier.
It was rescinded in May 2025, two days before it took effect. The country-rationing apparatus died on paper. Mexico is fully supported on the major AI services today.
No law anywhere requires government ID to use an AI chatbot right now. The claim 'soon you will need ID to use AI' is a forecast, not a fact.
But age-verification is spreading fast, and that is the real vector to watch - a Supreme Court ruling removed the brake, and the checks are moving to the app-store layer.
Every restriction works at ONE chokepoint: your dependence on a provider, chip, or model you do not own. Own the weights, run them local - and there is no gate left to be denied at.