There is no single global rulebook, and there will not be one soon. What exists instead is a set of distinct national bets about how to govern a technology that crosses borders by default. The European Union bet on one horizontal law that scales its demands to the risk of each use. China bet on a dense stack of targeted rules tied to content and security, enforced without a single statute to point to. The United Kingdom bet on its existing regulators and declined to write an AI act at all. The United States, for now, has bet on doing very little at the federal level while trying to stop its states from filling the gap.
The map is also no longer a Brussels-and-Washington story. Vietnam's AI Law, effective March 2026, is Southeast Asia's first binding comprehensive statute and reaches foreign providers directly. Taiwan put a framework act into force in January. Peru, which quietly passed Latin America's first AI law back in 2023, gave it real structure with a regulation that took effect this year. Australia looked at all of this and formally decided not to legislate. Each of those choices now sits in the register with the same fields and the same sourcing.
These choices reach well past their own borders. The EU AI Act binds any company whose systems are used in the Union, wherever that company sits, and its penalties run to 7 percent of worldwide turnover, the top of a wildly uneven global penalty table. South Korea's law reaches foreign providers serving Korean users, and Vietnam's does the same. A product built in one country now answers to the strictest regime it touches, which is why most global firms build to the European standard and treat the rest as a subtraction from it.
The American picture is the messiest and the fastest-moving. Congress has passed nothing. In its place, dozens of states have written their own measures, eleven of which are filed in the state tracker, and in December 2025 the White House signed an order directing the Justice Department to challenge them and threatening federal funding for states that keep them. Colorado, the first state to pass a comprehensive AI law, repealed it before it ever took effect. Virginia's legislature passed one and watched it die by veto. Until the courts settle the preemption fight, every state law on the books remains enforceable, and most follow the person affected rather than the company's address, so one national product can trigger several at once.
This is a working map of all of it. Every jurisdiction is filed the same way: the governing instrument named, the status and dates stated plainly, and a link to the primary text so you can check the wording yourself. The deadline calendar tracks every date that matters through 2028, and the glossary defines the terms of art the statutes lean on. It is free, and it is kept current, because currency is the whole point. A guide that still lists Colorado's original act as law, or the EU's high-risk deadline as August 2026, is worse than no guide at all.
