EU AI Act prohibition on emotion recognition: Digihumanism provides detailed feedback to Dutch supervisory authority

As coordinating supervisor on algorithms and AI, the Dutch data protection authority issued a second call for input concerning prohibited AI systems under the EU AI Act, this time regarding emotion recognition in the areas of workplace and education institutions.

⏳The prohibitions will apply as of February 2025.

⌛️ Supervisory authorities will start imposing sanctions for violation of the prohibitions in August 2025.

The feedback received will help the Dutch DPA:

👉 define their position with regard to their interpretation of these prohibitions for the purpose of supervision.

👉 dialogue with the Europe Commission in the drafting of guidelines on prohibited AI practices.

Karine Caunes, Digihumanism’s Executive Director stated: “Interpretation of the prohibited AI practices should be based on the very reasons why they are prohibited in the first place”

In the case of emotion recognition, are at stake:

⚠️ the accuracy of the AI system in recognising emotions or intentions

There is simply no scientific grounds underpinning the correlation or even cross correlation between certain biometric data, such as facial expressions, voices or brain activity and certain emotions.

There is no automaticity in establishing such links. It is not because in some cases, a given facial expression translates a certain emotion that this facial expression translates always the same emotion or that one correlation can be generalised and applied to all human beings.

The “inference” of emotions or intentions from biometric data is a pseudo-scientific misnomer and narrative to legitimate very dangerous practices.

Emotion AI signals the resurgence and widespread propagation of renewed pseudo-scientific phrenology or physiognonomy theories which pullulated during European darkest hours of colonialism and nazism.

⚠️ highly intrusive AI systems. They violate, without consent, the right to data protection and privacy, as well as the right to mental integrity and freedom of thought, right to liberty and security, all enshrined in the EU Charter of fundamental rights.

⚠️ discrimination against minorities such as Asian or Black people or neurodivergent people.

More specifically:

⚠️ in educational institutions, the use of such systems are contrary to the best interest of children, also guaranteed in the EU Charter of fundamental rights

⚠️ in “the areas of workplace”, the most widespread use relates to hybrid / telework. For the prohibition to have any meaning, it should apply to these situations: workplace is the place where one works.

🔥 To read our full submission, please click here.