Transparency Obligations for Providers and Deployers of Certain AI Systems

Providers and deployers must ensure users are informed when interacting with AI systems, unless it is obvious from the context. AI systems generating synthetic content (e.g., deepfakes) must clearly label outputs as artificially generated. Users must also be notified when AI is used for emotion recognition or biometric categorisation, except where required by law. Additionally, providers and deployers must disclose when AI systems create or alter content, unless exempt for legal, artistic, or satirical purposes. The AI Office will support the development of guidelines for detecting and labelling synthetic content.