AI systems that carry specific transparency risks, as defined by the EU AI Act. These systems are not banned or heavily regulated but must adhere to specific disclosure obligations, such as informing users they are interacting with an AI (e.g., chatbots) or labeling AI-generated content like deepfakes.