Definition
The Act distinguishes different AI risk categories and imposes stricter duties where AI can affect rights, safety, or important decisions.
Why it matters
Compliance screening must be purpose-specific, explainable, proportionate, and designed to avoid prohibited social scoring.