The European Parliament approved its negotiating position on the new legislation for artificial intelligence (AI) in the EU yesterday, in Strasbourg. The vote resulted in 499 in favor, 28 against, and 93 abstentions.

The Parliament will now enter into negotiations with the Council to reach an agreement on the final form of the legislation. The aim of the rules is to make AI human-centric, reliable, and free from harmful effects on health, safety, fundamental rights, and democracy.

These rules will ensure that AI technology is developed and used in Europe with full respect for the rights and values of the EU, such as human supervision, privacy, transparency, non-discrimination, and consideration for social and environmental issues.

The new rules will categorize AI systems based on the level of risk they represent and establish obligations for both providers and users within each category.

According to the draft legislation, AI systems classified as posing an excessively high risk to human safety will be completely prohibited, such as in cases of "social scoring."

The Members of the European Parliament have expanded this specific category to also include uses of this technology that violate privacy or lead to discrimination, such as real-time or retroactive facial biometric identification in public spaces, biometric categorization based on sensitive personal data (e.g., gender, race, ethnic origin, nationality, religion, political orientation), predictive policing (based on profiling, location, or past criminal behavior), emotion recognition for law enforcement, border control, workplaces, and educational institutions, and untargeted collection of facial images from the internet or closed-circuit surveillance cameras for the purpose of creating databases for facial recognition (in violation of human rights and the right to privacy).

The Members of the European Parliament have also ensured that the high-risk list includes AI systems that cause significant harm to health, safety, fundamental rights, and the environment.

Additionally, the list includes systems used to influence voters and thus election results, as well as content recommendation systems made by social media platforms that target over 45 million users based on algorithms.

Programming companies developing new types of AI applications, known as "foundation models," will be required to assess and mitigate potential risks (to health, safety, fundamental rights, the environment, democracy, and the rule of law) and register these applications in the EU database before making them available on the European market.

They must ensure that all productive AI applications based on such models, like ChatGPT, comply with the transparency requirements of the EU (by informing the public that they have been used to create an AI product) while also assisting in the detection of manipulated images. Moreover, they should receive appropriate human oversight to prevent their malicious use.

Moreover, they will be obligated to provide the public with a comprehensive and detailed record of the intellectual property-protected material used in their systems' "training."

To foster innovation and support small and medium-sized enterprises, European Parliament members propose exempting research activities and open-source AI software from these obligations.

Furthermore, the new legislation calls for the establishment of "regulatory sandboxes," as they are called, which will be implemented by public authorities in each country. These sandboxes will serve as real-life testing grounds for all AI applications before they are made accessible to the public.

Lastly, European Parliament members aim to strengthen citizens' rights to file complaints regarding AI systems and their entitlement to receive explanations for decisions made using high-risk systems that significantly impact their fundamental rights.

Following the vote, co-rapporteur Brando Benifei emphasized the significance of the moment, stating, "today, the world's attention is focused on us. While technological behemoths warn about the dangers posed by their own creations, Europe is taking proactive steps by proposing specific regulations to address the emerging risks associated with artificial intelligence."

He further added that "our objective is to nurture the positive potential of this transformative technology, enabling innovation and productivity, while simultaneously safeguarding our global standing and addressing the threats it poses to our democracies and freedoms in our negotiations with the Council."

Co-rapporteur Dragos Tudorache highlighted the importance of the legislative act on artificial intelligence, stating, "this legislation will serve as the foundation for global discourse on the development and governance of artificial intelligence. By ensuring that this powerful technology, poised to revolutionize our societies with its immense benefits, is developed and utilized in accordance with European values of democracy, fundamental rights, and the rule of law."