The Artificial Intelligence Act is a European Union regulation that establishes a common legal framework for AI across the EU1. It entered into force on 1 August 2024, with obligations phased in over roughly 6 to 36 months1.
The Act uses a risk-based approach to classify AI systems1. Unacceptable-risk uses are banned, high-risk systems are allowed only with strict requirements, limited-risk systems mainly face transparency duties, and minimal-risk systems are largely unregulated1.
The law primarily places obligations on AI providers and organizations that use AI professionally1. It also excludes some contexts such as military, national security, research, and non-professional applications1.
The Act includes special rules for general-purpose AI, including generative models such as ChatGPT1. These systems face transparency requirements, with lighter treatment for open-source models and additional evaluations for highly capable models1.
The Act also creates an European Artificial Intelligence Board to support coordination and compliance across member states1. It can apply extraterritorially when AI used in the EU is supplied by non-EU providers1.