Most workplace AI rules in the EU are about doing things carefully: inform staff, keep a person in charge, check for bias. A few uses are simply banned. The ban that matters most for employers is on AI that infers the emotions of workers and job candidates. It has applied since 2 February 2025, and the European Commission has published guidelines with concrete workplace examples. This page sets out what is banned, what is not, and the narrow exceptions.
The short answer: the EU AI Act prohibits AI systems that infer the emotions of people at the workplace, including candidates during hiring, except for medical or safety reasons (Article 5(1)(f)). It also bans biometric categorisation to infer sensitive traits such as trade union membership, and certain social scoring and manipulative practices. Breaches can bring fines of up to EUR 35 million or 7% of worldwide annual turnover, whichever is higher (Article 99(3)).
The ban on emotion recognition at work
Article 5(1)(f) prohibits placing on the market, putting into service for this specific purpose, or using AI systems to infer emotions of a natural person in the areas of workplace and education institutions, except where the system is intended for medical or safety reasons (Regulation (EU) 2024/1689). An emotion recognition system is an AI system for identifying or inferring emotions or intentions of people on the basis of their biometric data (Article 3(39)), which covers data from faces, voices and other physical or behavioural characteristics.
The reason is set out in recital 44: there are serious concerns about the scientific basis of such systems, since the expression of emotions varies across cultures, situations and individuals, and the imbalance of power at work combined with their intrusive nature could lead to unfair treatment.
What the Commission's guidelines say
The Commission's Guidelines on prohibited AI practices, C(2025) 5052 of 29 July 2025, are non-binding, but they show how the Commission reads the ban (Commission Guidelines on prohibited AI practices). On workplace emotion recognition they say:
- "Workplace" is broad. It covers any physical or virtual space where people carry out tasks for their employer or organisation, from offices, factories and warehouses to shops, open-air sites and cars, whatever the person's status: employee, contractor, trainee or volunteer. It also covers candidates during selection and hiring.
- Prohibited examples: a call centre using webcams and voice recognition to track employees' emotions such as anger; monitoring the emotional tone of hybrid work teams from video calls; emotion recognition during recruitment or during the probationary period; a supermarket using cameras to track employees' emotions such as happiness.
- Not prohibited by this article: a call centre tracking customers' emotions, for example to help staff cope with angry callers; detecting a frown or a smile without inferring an emotion; inferring physical states such as pain or fatigue, for example alerting a professional driver or pilot to rest.
- Personal training: emotion recognition used only for an employee's own training may be allowed if the results are not shared with HR, cannot affect assessment or promotion, and have no impact on the work relationship.
The medical and safety exception is narrow
The guidelines say the exception should be read narrowly. Therapeutic use means CE-marked medical devices. General monitoring of stress levels is not allowed, and an AI system to detect burnout or depression at work stays prohibited. Assessing employees' well-being, motivation or job satisfaction does not count as a medical reason. Safety means protecting life and health, not property against theft or fraud. An employer may measure stress or lack of concentration only where it poses a specific danger, for example around dangerous machines or chemicals, and may not then use the data for other purposes such as judging performance. Any such use must be limited to what is strictly necessary, with safeguards, and the guidelines point out that EU employment law consultation duties and data protection law still apply.
Other bans that can reach the workplace
- Biometric categorisation (Article 5(1)(g)): systems that categorise individuals based on their biometric data to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation.
- Social scoring (Article 5(1)(c)): evaluating or classifying people over time based on social behaviour or personal characteristics, where the score leads to detrimental treatment in unrelated contexts, or treatment that is unjustified or disproportionate. The guidelines note that specific employee evaluations are not banned as such if lawful and in line with the AI Act; those are generally high-risk instead.
- Manipulation and exploitation (Article 5(1)(a) and (b)): AI using subliminal, manipulative or deceptive techniques, or exploiting vulnerabilities due to age, disability or a specific social or economic situation, in a way that materially distorts behaviour and causes or is likely to cause significant harm.
Banned versus high-risk
Most workplace AI is not banned but high-risk: systems for recruitment, decisions on terms, promotion or termination, task allocation based on behaviour or traits, and monitoring and evaluating performance (Annex III, point 4). Those duties apply from 2 December 2027. Emotion recognition outside the workplace and education is not banned by Article 5(1)(f) but is listed as high-risk (Annex III, point 1(c)). The Act also lets the EU and Member States keep or introduce laws more favourable to workers on employers' use of AI (Article 2(11)). See AI at work and workers' rights and AI monitoring at work.
Practical steps
- Employers: check every tool that analyses voice, video, faces or wearables for features that infer mood, stress, engagement or sentiment of staff or candidates, including features inside call centre, video interview and collaboration software. Switch them off unless a documented medical or safety reason applies.
- Workers and candidates: you can ask whether any system analyses your emotions. If it does, raise it with your representatives or the national authority; see AI and whistleblowing at work.
This page is general information about EU rules, not legal advice.
Questions and answers
Is emotion recognition of employees banned in the EU?
Yes. Since 2 February 2025, Article 5(1)(f) of the AI Act prohibits AI systems that infer the emotions of people at the workplace, including job candidates, except for medical or safety reasons.
Can a call centre use AI to analyse emotions?
Not its employees' emotions. The Commission's guidelines give that as a prohibited example. Analysing customers' emotions is not covered by this ban, though other rules such as data protection still apply.
Is AI that detects worker fatigue allowed?
Fatigue and pain are physical states, not emotions, so systems that detect them, for example to alert a driver to rest, fall outside the ban. Other rules, such as data protection, still apply.
Can employers use emotion AI to check staff well-being or burnout?
No. The guidelines say assessing well-being, motivation or job satisfaction is not a medical reason, and a system to detect burnout or depression at work remains prohibited.
What are the penalties?
Breaching the Article 5 prohibitions can lead to fines of up to EUR 35 million or 7% of total worldwide annual turnover, whichever is higher.