US and EU AI rules point to need for global framework

The US and EU have taken sharply different approaches to AI governance. The EU AI Act, in force since 1 August 2024, establishes a comprehensive risk-based regime covering banned practices, high-risk systems, limited-risk transparency duties and general-purpose models. The US still lacks a comprehensive federal statute, relying instead on executive action, voluntary NIST guidance and state-level laws.

Recent US developments show both experimentation and instability. A bipartisan draft of the Great American AI Act emerged in June 2026, while state measures in Texas, Colorado and California have pursued consumer-protection, disclosure and frontier-model safety obligations. Federal pressure against state regulation and industry challenges have weakened some efforts, including Colorado’s broader AI law before it took effect.

The EU model offers clearer rights-based governance, but its complexity has prompted delays under the Digital Omnibus, with high-risk obligations deferred to December 2027 and August 2028. A global framework would aim to settle which AI uses should be banned outright, including fully autonomous weapons, mass surveillance, nudification apps and AI-delivered mental health care, while applying risk-tiered regulation to the remaining systems.