G20 backs the Carolina Principles while the EU tightens its AI Act

On September 1 in Chapel Hill, North Carolina, the United States secured unanimous backing from G20 ministers for a light-touch AI governance framework. The same week, the European Union kept actively enforcing the high-risk obligations of its AI Act. The world isn't converging on a single rulebook for AI — it's splitting into two.

What the Carolina Principles actually are

At the G20's Innovation Ministerial, White House science and technology advisor Michael Kratsios presented what the US administration is calling the "Carolina Principles": a framework urging G20 member countries to take an "innovation-first," minimal-intervention approach to governing artificial intelligence and other emerging technologies. Representatives from the world's largest economies unanimously agreed to back these guidelines, which explicitly steer away from creating new AI-specific regulatory bodies.

The EU is heading the opposite direction

While Washington was pushing this approach in Chapel Hill, the EU's AI Act reached its most consequential enforcement milestone yet: as of August 2, 2026, requirements for "high-risk" AI systems — those used in hiring, credit decisions, access to education, and law enforcement — are now legally binding. The EU's AI Office has already sent formal compliance letters to more than thirty frontier AI labs, a move that confirms Brussels has no intention of softening its timeline under pressure from Washington.

What the regulatory split means for a company in Latin America

For a Guatemalan or Latin American company selling AI-powered software to customers in Europe — or simply using models from vendors that operate in the EU — the AI Act isn't a distant concern: if the AI system is used for hiring, credit, or similar decisions and touches European data or users, the high-risk obligations apply exactly as they would to a company headquartered in Berlin. At the same time, the US stance at the G20 suggests the American market will likely stay comparatively more permissive in the short term, which can create the temptation to build a single product "for the more lenient market" and defer European compliance until a fine arrives.