Behind the scenes of the AI Act: How will the new regulations change the world?

Biometric data is used for tracking citizens and detecting emotions, while social scoring operates within the social credit system, classifying people based on their social behavior, socio-economic status, and personal traits. Additionally, criminal prediction involves profiling individuals who may potentially commit crimes. These artificial intelligence tools have been successfully employed by the Chinese authorities to control their citizens. However, in the European Union, they will be banned thanks to the AI Act – a comprehensive legal regulation concerning the operation and use of artificial intelligence. The AI Act is the first of its kind in the world.

Efforts in the field of artificial intelligence development have been ongoing since the 1950s. In recent years, and even months, the progress in this area has significantly accelerated. Generative AI and cloud-based solutions are utilized in various sectors of the economy. Artificial intelligence has become a general-purpose technology, applicable in every sector and aspect of our lives. Therefore, it is crucial and urgent to establish a legal framework for its functioning. The approach to this issue varies greatly worldwide, as we have previously discussed in the article „What’s happening with AI? Will the European Union Regulate the Operation of Artificial Intelligence?”

For the European Union, technological development must be in agreement with the European values, written in the Charter of Fundamental Rights.

The proposal for horizontal regulation represents pioneering actions. In the process of a lengthy legislative process, on June 14, 2023, the European Parliament approved the AI Act project in its official position.

Trilateral negotiations on the final shape of the text between the European Parliament, the European Commission, and the member states are already ongoing. There is a chance that the AI Act will be adopted before the next European Parliament elections scheduled for May 2024.

There is great determination and willingness among EU bodies to adopt the legislation by the end of the current term. The most efficient politicians have been appointed to report on the efforts.

The European Union does not aim to regulate all AI-based systems but rather only those that are considered to be risky. The principle of risk assessment using a so-called risk-based approach has been a guiding principle in EU legislative processes for several years. This principle has also been applied in this case.

The definition of high-risk areas has been expanded to include harm to human health, safety, fundamental rights, or the environment. AI providers will be required to ensure the protection of fundamental rights, health and safety, the environment, democracy, and the rule of law.

The European Union does not want to regulate all artificial intelligence systems, only those that pose risks. The risk-based approach has been the leading principle in the legislative processes at the European Union level for several years. It has also been applied in this case.

The classification of high-risk areas has been expanded to include harm to human health, safety, fundamental rights, or the environment. AI providers will be required to guarantee the protection of fundamental rights, health and safety, the environment, democracy, and the rule of law.

Artificial Intelligence systems with an unacceptable level of risk to human safety will be prohibited, including the use of subliminal and manipulative techniques, or social scoring used for community policing. Real-time biometric identification will also be banned, which is a crucial step in protecting freedom of speech, freedom of assembly, and non-discrimination in public spaces. The Parliament has also voted in favor of banning biometric categorization based on sensitive features such as gender, race, sexuality, or ethnic origin, as well as the prohibition of emotion recognition in education, in the workplace, by the police, and at the borders (including in the field of criminal prediction).

Brando Benifei (Member of the Progressive Alliance of Socialists and Democrats in the European Parliament, co-rapporteur of the AI Act) points out:

„As the European Parliament, we will support the amendment (…) extending the ban on social scoring to private entities. This would prohibit the use of artificial intelligence to classify citizens for illegal purposes, as is the case in other parts of the world. (…) One thing we don’t want to change, which is already in the draft law, is the ban on real-time biometric recognition. We will look at exceptions for law enforcement, but we do not want this practice to be generally accepted because it can create many problems in terms of controlling citizens’ lives. (…) The fact that artificial intelligence covers various sectors, and above all, that it is not yet fully understood, is one of the biggest challenges. (…) Legislation that is flexible in terms of updates and can be adapted as things change will play a crucial role because AI is still a rapidly evolving field.”

Despite an attempt by the European People’s Party to reverse the agreement on biometric surveillance, Members of the European Parliament have shown that they listen to the voices of citizens (e.g., the Reclaim Your Face campaign) who expect public spaces free from facial recognition and other mass biometric monitoring systems.

The AI Act also emphasizes the integrative role of the new AI Agency. Furthermore, there is a proposal for closer alignment with the General Data Protection Regulation (GDPR). In their amendments to the Commission’s proposal, it was indicated that artificial intelligence systems should be safe, transparent, human-supervised, identifiable, non-discriminatory, and environmentally friendly.

Unfortunately, the European Parliament did not introduce provisions that would protect the rights of migrants from discriminatory surveillance systems, which could facilitate illegal deportations or discriminatory profiling.

Despite leaving some loopholes in the AI Act regulations, the new law is a significant step forward in protecting our fundamental rights. It provides hope for better regulation and responsible use of artificial intelligence, paving the way for further improvements and adjustments in the future.

Marta Stępniak

Translated from Polish by Agnieszka Nikodem, July 2023