LEGAL PROHIBITION OF EMOTION RECOGNITION SYSTEMS IN THE WORKPLACE

This article examines the legal prohibition of emotion-recognition systems in the workplace as a proportionate response to unresolved scientific and ethical defects of the technology. Systems that infer inner affect from biometric cues—facial movements, voice, gaze, or physiological signals—presuppose a stable mapping between expression and emotion that decades of research have not validated and that varies across cultures and individuals. Against this backdrop of unreliability and structural power asymmetries in employment, the paper analyses the EU AI Act’s risk-based architecture and its outright ban under Article 5(1)(f) in work and education, subject only to narrow medical and safety exceptions. It contrasts the European model with national measures (notice, bias auditing, transparency), details timelines and penalties, and evaluates downstream effects on hiring, performance management, and worker well-being. Building on evidence of bias and weak construct validity, the article advances policy proposals: recognising affective data as a special sensitive category, instituting public registries of independent audits, extending prohibitions to other high-risk contexts, and deploying human-centred alternatives such as participatory design, ethics oversight, and AI-literacy programmes. The contribution lies in demonstrating why prohibition is justified absent reliable science and in offering a practicable compliance toolkit that safeguards worker autonomy while allowing beneficial, narrowly tailored uses of AI.