Emotion Recognition and the AI Act Regulation: The New Restrictions and Implications!
European Commission publishes Artificial Intelligence (AI) Prohibited Practices Guidelines under the AI Act: On February 4, 2025, the European Commission published AI Prohibited Practices Guidelines under the AI Act. These guidelines provide a clear view of practices deemed unacceptable due to risks to European values and fundamental rights, focusing on malicious manipulation, social scoring and real-time remote biometric identification.
The AI Act classifies AI systems into different risk categories, including prohibited risks, high risks and those subject to transparency obligations. The guidelines aim to ensure uniform application of the AI Act across the European Union by providing legal explanations and practical examples, but final interpretations are reserved for the Court of Justice of the European Union (CJEU). At the same time, these guidelines underline the European Union's commitment to promote a safe and ethical AI landscape, supporting innovation but guaranteeing the protection of health, safety and fundamental rights.
A controversial aspect of the AI Act is the restrictions placed on emotion recognition technologies, which have been banned in educational institutions and the workplace, except for medical or safety purposes. This decision was taken without proper consultation with the experts in the field, which generated criticism from the academic community and industry. In response to these concerns, the Office of Artificial Intelligence developed guidance to clarify the bans, a document published on February 4, 2025, just two days after the restrictions came into effect.
The main prohibitions on the recognition of emotions
1. Prohibited domains, AI systems that recognize emotions are prohibited in:
• Workplaces – employers cannot use such systems to analyze the emotional state of employees, except for safety reasons.
• Educational institutions – schools and universities cannot implement such technologies to monitor the attention or interest of pupils/students.
2. What is allowed? There are some notable exceptions:
• Medical – emotion recognition is allowed if it is used for therapeutic purposes or to assist people with disabilities, such as helping autistic patients or people with visual or hearing impairments.
• Public safety – systems that detect emotions can be used in situations involving the protection of life or health, such as monitoring the stress of drivers to prevent accidents.
3. What is considered emotion recognition? The guidance clarifies that the prohibition includes not only direct detection of emotions, but also inferences based on gestures, tone of voice or facial expressions. However, simple observations of visible expressions (eg a smile or a frown) without the use of AI are not considered emotion recognition.
4. Implications for research and industry. The AI Act allows the use of emotion recognition technologies for research purposes as long as they are not implemented in a real work or educational environment. In addition, AI systems that analyze written text to detect sentiment (for example, sentiment analysis in online reviews) are not subject to the bans.
5. Penalties for Failure to Comply. Starting August 2, 2025, penalties will be applied for non-compliance with the restrictions imposed by the AI Act. AI providers must also include in their product terms and conditions a ban on the use of these technologies in illegal contexts.
New European regulations pose significant challenges for developers and users of AI technologies in the field of emotion recognition. Although there are exceptions for medical and safety purposes, strict bans in workplaces and educational institutions will have a major impact on commercial applications and research. It remains to be seen how these restrictions will affect the EU's competitiveness in relation to other regions, such as the US and China, which continue to advance rapidly in AI.
Source:
https://digital-strategy.ec.europa.eu/en/library/commission-publishes-guidelines-prohibited-artificial-intelligence-ai-practices-defined-ai-act
Share:
Comments