South Korea AI Basic Act enforces first full AI law

South Korea’s AI Basic Act is the first end-to-end, fully in-force AI statute, and that makes it a global reference point for how governments might govern increasingly capable systems. Unlike narrow deepfake or sectoral rules, it treats AI as critical infrastructure: high‑impact systems must have meaningful human oversight, all AI-generated content must be labeled, and regulators gain real enforcement tools. That combination of teeth and clarity will shape how both domestic and foreign firms design, deploy, and document advanced models.

For the race to AGI, Korea just turned regulatory uncertainty into a competitive lever. Multinationals now have to ask whether they build to the strictest regime (Korea and, soon, the EU) and roll that compliance out globally. If Korean enforcement proves predictable and not overly punitive, the country could become a preferred testbed for safety‑critical deployments, particularly in finance, health and critical infrastructure. But if compliance becomes box‑ticking or politicized, it could push cutting‑edge work elsewhere.

Either way, the law accelerates the shift from voluntary ‘AI principles’ to binding governance for systems that will look more and more like proto‑AGI. Companies aiming at general-purpose, agentic models now have to assume that transparency, labeling and human‑in‑the‑loop requirements will be part of the operating environment, not optional extras.