The potential and risks of using AI agents
Print
AI agents combine large language models with autonomous decision-making capabilities, enabling them to perform tasks, access information, and interact with digital systems with limited human supervision. This TA Compact study examines their opportunities and risks in three domains: the workplace, consumer-oriented services, and democratic discourse. It analyses technical characteristics, key use cases, and challenges related to autonomy, privacy, security, consumer protection, liability, and manipulation. The project aims to identify emerging policy issues and develop options for evidence-based parliamentary action.