Guidelines on prohibited artificial intelligence practices established by Regulation (EU) 2024/1689 (AI Act)

Key Findings

• — Provides a worked example of a therapeutic chatbot offering mental-health support and coping strategies that exploits users' limited intellectual capacities to nudge them toward buying expensive products or behaving in harmful ways, illustrating Article 5(1)(b) (exploitation of vulnerabilities)
• — Discusses AI companionship applications that use anthropomorphic features and emotional cues to influence users' feelings and dispositions as a potential Article 5(1)(a) manipulation concern, while noting companionship systems designed for engagement without manipulative or deceptive practices generally fall outside the prohibition
• — Frames psychological harm from manipulative AI systems as encompassing adverse effects on mental health and emotional wellbeing, noting such harms can accumulate over time and be difficult to measure
• — Provides a further example of AI systems identifying and targeting women and girls with disabilities for exploitative grooming

Methodology Notes

European Commission guidance document (non-binding interpretive guidance; final interpretive authority rests with the Court of Justice of the EU), 134 pages, adopted in Brussels 29 July 2025. Verified via direct download and full-text extraction of the primary PDF.

Sources