EU rules tighten on emotion AI
- Brussels is moving a specific deadline that will classify customer‑emotion AI as high‑risk this August 2026. - The change directly targets contact centres and emotion‑analysis systems used on customers. - Companies must treat these classifiers as high‑risk under the EU AI Act, forcing compliance work on data, logging and legal basis now (cxtoday.com).
The European Union’s main AI law hits contact centres next on 2 August 2026, when customer emotion-analysis systems start falling under its high-risk rules. (digital-strategy.ec.europa.eu) The Artificial Intelligence Act entered into force on 1 August 2024, banned certain uses from 2 February 2025, and applies most Annex III high-risk obligations from 2 August 2026. The European Commission’s implementation timeline lists that August 2026 date as the point when Annex III high-risk rules and enforcement start. (digital-strategy.ec.europa.eu) (ai-act-service-desk.ec.europa.eu) Emotion-recognition AI tries to infer how a person feels from signals such as voice, facial expression, keystrokes, or behavior. The Act already bans emotion-recognition systems in workplaces and schools, with narrow medical or safety exceptions, but customer-facing uses are treated differently. (digital-strategy.ec.europa.eu 1) (digital-strategy.ec.europa.eu 2) For customer service, the legal shift comes through Annex III, the part of the law that lists stand-alone high-risk uses. Systems in that bucket face mandatory controls on risk management, data quality, technical documentation, record-keeping, human oversight, accuracy, robustness, and cybersecurity. (artificialintelligenceact.eu) (eur-lex.europa.eu) The compliance load is not just for developers. Providers of high-risk systems must keep a quality-management system, complete conformity assessments, and register covered systems in the European Union database, while some deployers must carry out a fundamental-rights impact assessment before use. (artificialintelligenceact.eu) (eur-lex.europa.eu) (licentium.io) That matters for contact-centre software because many products market “sentiment,” “empathy,” or “emotion” scoring as a way to steer agents, route calls, or flag unhappy customers in real time. Outside the workplace ban, those customer-facing systems now sit in the part of the law that triggers formal audits, logging, and governance work before August. (cxtoday.com) (eur-lex.europa.eu) The Commission is also trying to soften the rollout. Its AI Act Service Desk says a “Digital Omnibus” proposal would link the application of high-risk rules to the availability of support tools, including harmonised standards, but that proposal has not replaced the current 2 August 2026 date. (ai-act-service-desk.ec.europa.eu) For companies selling or using customer emotion AI in Europe, the calendar is now concrete: prohibited workplace uses are already off-limits, and customer-facing systems are heading into the high-risk regime on 2 August 2026. (digital-strategy.ec.europa.eu) (ai-act-service-desk.ec.europa.eu)
