The EU AI Act bans a specific set of “unacceptable‑risk” AI practices—uses considered so harmful to fundamental rights, safety, or democratic values that they are prohibited outright. These include manipulative systems, exploitation of vulnerable groups, social scoring, crime‑prediction based on profiling, untargeted facial scraping, and emotion‑recognition in schools and workplaces. Below is a clear, structured breakdown of every prohibited use, with citations to the legal text.
🚫 Prohibited AI Uses Under the EU AI Act (Article 5)
The EU AI Act’s Article 5 lists unacceptable‑risk AI systems that cannot be placed on the market, put into service, or used in the EU. These prohibitions apply to both EU and non‑EU companies offering AI in the EU.
- Subliminal, Manipulative, or Deceptive AI Techniques
AI systems that use subliminal techniques or purposefully manipulative/deceptive methods to distort a person’s behavior in a way that impairs their ability to make informed decisions and causes (or is likely to cause) significant harm.
- Exploitation of Vulnerable Groups
AI systems that exploit vulnerabilities due to age, disability, or social/economic situation to distort behavior in a way that causes (or is likely to cause) significant harm.
- Social Scoring Systems
AI systems that evaluate or classify people based on social behavior or personal traits, leading to:
- Unfavorable treatment in unrelated contexts, or
- Unjustified or disproportionate harm.
This is similar to China‑style social credit scoring and is banned outright.
- Predictive Policing Based on Profiling
AI systems that assess or predict the likelihood of a person committing a crime based solely on profiling or personality traits, without objective, verifiable facts.
(Important: systems that support human assessment based on real evidence are not banned.)
- Untargeted Facial Recognition Scraping
AI systems that create or expand facial recognition databases by scraping images from the internet or CCTV footage without consent.
This targets companies that mass‑harvest biometric data.
- Emotion Recognition in Schools and Workplaces
AI systems used to infer emotions of individuals in:
- Workplaces, or
- Educational institutions,
except for medical or safety purposes.
This includes classroom monitoring tools and workplace productivity based on emotion surveillance.
The EU AI Act imposes a strict, outright ban on emotion‑recognition AI in schools and workplaces because regulators concluded the technology is unreliable, intrusive, and fundamentally incompatible with the power dynamics in classrooms and employment settings.
🎭 What the Ban Actually Covers
Under Article 5(1)(f) of the EU AI Act, it is illegal to place on the market, deploy, or use any AI system intended to detect or infer emotions of:
- Students in educational settings
- Employees in workplace settings
This includes systems that analyze:
- Facial micro‑expressions
- Voice tone or stress levels
- Body posture or gestures
- Physiological signals (heart rate, skin response, breathing)
- Brain‑activity‑based signals (e.g., EEG headsets)
If the system can infer emotions from biometric data, it is banned.
🧠 Why the EU Banned Emotion Recognition in These Settings
- The technology is scientifically unreliable
Experts and regulators note that emotion‑recognition systems often cannot accurately infer internal emotional states from external signals.
- High risk of discrimination and unfair treatment
Incorrect emotion readings could lead to:
- Biased hiring decisions
- Wrongful discipline
- Misinterpretation of student behavior
- Power imbalance makes consent meaningless
Employees and students cannot freely refuse such monitoring.
- Threats to privacy and human dignity
Emotion monitoring is considered deeply intrusive, especially when tied to biometric data.
🏫 Potential Examples of Banned Uses in Schools
- Classroom cameras analyzing student engagement, if based on biometric analysis
- AI tools rating student “attention,” “motivation,” or “stress”
- Systems monitoring emotional states during exams
- Emotion‑tracking wearables used in school programs
Even if marketed as “improving learning,” these systems are prohibited.
🏢 Potential Examples of Banned Uses in Workplaces
- AI tools evaluating job applicants’ emotions during interviews
- Call‑center software judging whether employees sound “friendly enough”
- Wearables tracking stress, fatigue, or emotional tone
- Monitoring employee mood to predict burnout or productivity
These examples match real‑world deployments regulators explicitly cited.
🚨 The Only Two Exceptions
The ban includes two narrow exceptions, allowed only when emotion recognition is used for:
- Medical purposes
For example:
- Clinical mental‑health assessments
- Assistive technologies for neurological conditions
- Safety purposes
For example:
- Detecting driver fatigue in industrial vehicles
- Monitoring emotional distress in high‑risk environments
Outside these exceptions, the ban is absolute.
⚖️ Penalties for Violations
Using emotion‑recognition AI in schools or workplaces can trigger the highest tier of fines under the EU AI Act:
- Up to €35 million or
- 7% of global annual turnover
Whichever is higher.