Digital, Entertainment & Sports
The European Parliament approves the European Union Regulation on the regulation of Artificial Intelligence
Patricia Lalanda, Fernando Martín ·
The recent step taken by the European Parliament in approving the Artificial Intelligence (“AI”) Regulation on 13 March 2024 marks a significant milestone in the legal and technological spheres. This regulation, designed to regulate the development and use of artificial intelligence in the European Union (“EU”), presents a number of key legal aspects that must be considered.
- Key legal aspects and objectives:
- Protection of fundamental rights: The regulation establishes safeguards to ensure that artificial intelligence respects human rights, including privacy, non-discrimination and equality.
- Legal liability: The concept of operators and providers of artificial intelligence systems is introduced, who will be responsible for ensuring compliance with the provisions of the regulation and will face penalties in the event of non-compliance.
- Risk assessment: A framework is established for assessing and classifying the risk associated with artificial intelligence systems, which will determine the level of compliance and the corresponding legal obligations.
- Transparency and traceability: Artificial intelligence systems are required to be transparent in their operation and capable of providing explanations for the decisions they make.
- The Regulation also emphasises the European Union’s intention to balance fundamental rights with compatibility with and the promotion of technological innovation in the EU.
- Risk levels and legal requirements:
- Unacceptable AI: These will be AI systems that violate the EU’s fundamental rights and pose a direct threat to public safety and privacy. Their use is prohibited, except by law enforcement authorities in cases involving national security.
- High risk: Artificial intelligence systems considered high-risk will be subject to stricter requirements, such as comprehensive risk assessments, mandatory certifications and ongoing supervision. Example: AI systems used in critical sectors: This includes systems used in healthcare, transport, education and public safety, among others. These systems may directly affect people’s safety and well-being, and are therefore subject to greater regulatory attention.
- AI systems for biometric purposes: Facial recognition systems and other forms of biometric identification are considered high-risk due to their ability to collect and process sensitive personal data.
- AI systems affecting children’s rights: This covers systems that collect data from minors or have a significant impact on their well-being, education or development.
- AI systems used in judicial or administrative proceedings: Law enforcement and the administration of justice are sensitive areas where the accuracy and impartiality of AI systems are critical to ensuring fairness and justice.
These are some examples of the types of artificial intelligence systems considered high-risk under the European Union’s Artificial Intelligence Regulation. It is important to emphasise that the regulation may include additional provisions or adjustments as it is implemented and adapted to changing technologies and emerging risks.
- Limited risk: The most common systems. They must comply with certain specific obligations concerning transparency and warnings about the use of AI, and add a “Disclaimer” warning that the content has been generated using AI.
- Minimal or no risk: Proportionate measures will apply to low-risk and minimal-risk artificial intelligence systems, with a focus on minimising risks and protecting individuals’ rights. They are not specifically regulated. Citizens may freely decide whether to use them. For example, a video game using AI or spam filters.
- Timetable for entry into force and legal implications.
The regulation will enter into force 20 days after its publication in the Official Journal of the European Union and will, as we indicated, be fully applicable two years later, although there are some exceptions that will enter into force before that period expires:
- Titles I and II, general provisions and prohibited AI practices, including AI considered unacceptable, will enter into force 6 months later.
- The provisions of Chapter III, Section 4 (notifying authorities and notified bodies), Chapter V (general-purpose AI models), Chapter VII (governance) and Chapter XII (penalties) will apply twelve months after entry into force, with the exception of Article 101.
- Article 6(1), high-risk AI systems, and the corresponding obligations will apply 36 months after its entry into force.
At Loyra Abogados, as a law firm with 40 years of experience advising on New Technologies, we are committed to helping our clients understand and comply with evolving artificial intelligence regulations in the European Union and worldwide.
LOYRA ABOGADOS
New Technologies Department