Table of contents
- AI practices prohibited by the new European regulation
- The definition of prohibited AI practices
- The risks of emotion recognition and biometrics
- Remote biometric identification systems
- Implications for AI system providers
AI practices prohibited by the new European regulation
The EU Regulation 2024/1689 (AI Act), adopted on July 12, 2024, represents a crucial step in regulating artificial intelligence systems within the European Union. This regulatory framework, also known as the AI Act, establishes ethical and operational boundaries to ensure safety, respect for fundamental rights, and protection against the misuse of technology.
The regulation will come fully into force in August 2026, with some key provisions taking effect from February 2025. Below, we analyze the prohibited AI practices and their impact.
The definition of prohibited AI practices
The regulation identifies prohibited artificial intelligence practices as those applications that pose an unacceptable risk to the safety, dignity, and rights of individuals.
These prohibitions, specified in Article 5 of the regulatory text, cover several critical areas, including:
- The use of subliminal or manipulative techniques that influence behavior without the user’s awareness. These systems, designed to exploit cognitive vulnerabilities, threaten the ability to make informed decisions.
- Social scoring systems based on AI that analyze behaviors or personal characteristics to assign social scores. This type of categorization, often used in unrelated contexts, can lead to discriminatory treatment.
- The use of predictive technologies to identify potential crimes based solely on psychological or behavioral profiles, raising significant privacy and individual freedom concerns.
The risks of emotion recognition and biometrics
One of the most controversial aspects involves the use of emotion recognition systems and advanced biometric technologies.
While these tools are often associated with significant progress in areas like security or healthcare, their use in public spaces or workplaces raises serious ethical questions. For instance, the regulation prohibits:
- The use of AI systems to infer emotions in contexts such as schools and offices, except for medical or safety reasons.
- The categorization of individuals based on biometric data to deduce sensitive information such as sexual orientation, political opinions, or religious affiliation.
These prohibitions aim to prevent discrimination and abuse, safeguarding privacy rights and ensuring technological neutrality.

Remote biometric identification systems
Another critical area involves real-time remote biometric identification systems, often used in law enforcement activities. The regulation allows their use only in extremely limited cases, such as:
- Searching for victims of trafficking or missing persons.
- Preventing imminent terrorist threats.
- Locating suspects linked to serious crimes.
Even in these contexts, member states must adhere to strict conditions, including obtaining prior authorization from an independent authority and ensuring proportional protections for human rights.
Implications for AI system providers
AI system providers are directly involved in ensuring compliance with the regulation. In addition to conducting fundamental rights impact assessments, they are required to register their systems in the EU database.
Non-compliance with these rules carries significant penalties, including fines of up to 6% of the company’s global turnover.
Questions and answers
- What are the prohibited AI practices under the European regulation?
These are artificial intelligence applications that pose unacceptable risks to safety and human rights, such as manipulative techniques or discriminatory categorizations.
- When will the EU regulation on artificial intelligence come into force?
The regulation will be fully operational in August 2026, with some provisions active from February 2025.
- What are the risks of emotion recognition?
These systems can violate privacy and lead to discrimination, particularly in sensitive contexts like schools or workplaces.
- What does the regulation say about biometric identification systems?
It bans their indiscriminate use in public spaces, except for exceptions related to security and crime prevention.
- What is the main goal of EU Regulation 2024/1689?
To ensure the safe, ethical, and respectful use of artificial intelligence systems in line with fundamental rights.
- What are the penalties for violating the regulation?
Companies can be fined up to 6% of their global turnover for serious violations.
- How does the regulation protect fundamental rights?
By prohibiting manipulative, discriminatory, or privacy-invasive AI practices.
- What does the AI Act mean by an unacceptable risk?
A level of risk that cannot be mitigated and compromises safety or human dignity.
- Are social scoring systems prohibited?
Yes, the regulation explicitly bans social scoring based on personal data.
- Which countries are subject to the regulation?
All member states of the European Union.