The Parliament Passed a Revolutionary Regulation on Artificial Intelligence

On Wednesday, Members of the European Parliament approved the Artificial Intelligence Act, which guarantees safety and protection of fundamental rights while simultaneously promoting innovation. The regulation, agreed upon during negotiations with member states in December 2023, was adopted with 523 votes in favor, 46 against, and 49 abstentions. Its aim is to protect human rights, democracy, the rule of law, and environmental sustainability from high-risk artificial intelligence, while also encouraging innovation and ensuring Europe’s role as a leader in this field. The rules for artificial intelligence are established based on its potential risks and level of impact.

The new rules prohibit certain applications of artificial intelligence that threaten citizens’ rights, including systems for biometric categorization based on sensitive characteristics and indiscriminate collection of facial images from the internet or surveillance footage for the creation of recognition databases. It also prohibits emotion recognition in workplaces and schools, social scoring of citizens, predictive policing (when it is based solely on profiling individuals or assessing their characteristics), and artificial intelligence that influences human behavior or exploits individuals’ vulnerabilities.

The use of remote biometric identification systems for criminal prosecution is generally prohibited, except in circumstances that are exhaustively listed and precisely defined. Real-time remote biometric identification may only be used if strict requirements are met, e.g., its use is time- and geographically limited, and a special judicial or administrative authorization has been previously issued for it. Some examples of such circumstances include targeted searches for missing persons or preventing terrorist attacks. Subsequent use of remote biometric identification systems is considered high-risk use and requires judicial authorization related to a criminal offense.

Clear obligations are also foreseen for other high-risk artificial intelligence systems (due to their significant potential to endanger health, safety, fundamental rights, the environment, democracy, and the rule of law). High-risk artificial intelligence is used, among other things, for critical infrastructure, education and training, employment, key private and public services (e.g., healthcare, banking), certain systems within criminal prosecution, migration and border management, justice, and democratic processes (e.g., influencing elections). For such systems, risk assessment and mitigation, record-keeping of use, transparency and accuracy, and human oversight must be ensured. Citizens will have the right to file lawsuits related to artificial intelligence and receive explanations for decisions based on high-risk artificial intelligence systems that affect their rights.

General-purpose artificial intelligence systems and the models they are based on must meet certain transparency criteria, including compliance with EU copyright law and the publication of detailed summaries of the content used for training. More powerful general-purpose artificial intelligence systems that could pose systemic risks will be subject to additional requirements, including model assessment, risk evaluation and mitigation, and incident reporting. Additionally, artificially created or manipulated audio content, photographs, and video content (deepfakes) must be clearly labeled as such.

– We finally have the world’s first binding law on artificial intelligence, aimed at reducing risks, creating opportunities, combating discrimination, and introducing transparency. Thanks to the Parliament, unacceptable practices of artificial intelligence will be banned in Europe, and the rights of workers and citizens will be protected. An Artificial Intelligence Office will now be established to support companies in complying with the rules before they come into force. We have ensured that human beings and European values are at the very center of artificial intelligence development,” said Brando Benifei (S&D, Italy), co-rapporteur of the Internal Market Committee, during the plenary debate on Tuesday.

– The EU has succeeded. We have woven the concept of artificial intelligence into the core values that form the foundation of our societies. However, much work remains that goes beyond the Artificial Intelligence Act itself. Artificial intelligence will prompt us to rethink the social contract that is at the heart of our democracies, our educational models, labor markets, and the way we conduct war. The Artificial Intelligence Act is a starting point for a new governance model built around technology. Now we must focus on implementing this law in practice,” said Dragos Tudorache (Renew, Romania), co-rapporteur of the Civil Liberties Committee.

A regulatory sandbox and real-world testing environment should be established at the national level, allowing startups and small and medium-sized enterprises access to develop and train innovative artificial intelligence systems before they are brought to market.

The regulation still needs to be reviewed by legal experts and linguists, and it is expected to be adopted before the end of the parliamentary term (in the so-called correction procedure). The Council also needs to officially adopt it. It will enter into force twenty days after publication in the Official Journal of the European Union and will start to apply 24 months after its entry into force, except for prohibited applications (six months after entry into force), codes of practice (nine months after entry into force), rules for general-purpose artificial intelligence including governance (twelve months after entry into force), and obligations for high-risk systems (thirty-six months after entry into force).