EU lawmakers approve AI Act

People visit the Mobile World Congress (MWC) 2024 in Barcelona, Spain, Feb. 26, 2024. (Xinhua/Gao Jing)

The European Parliament on Wednesday approved the Artificial Intelligence (AI) Act with an overwhelming majority in Strasbourg, France.

BRUSSELS, March 13 (Xinhua) -- The European Parliament on Wednesday approved the Artificial Intelligence (AI) Act with an overwhelming majority in Strasbourg, France.

The Parliament passed the act with 523 lawmakers voting in favor, 46 against and 49 abstentions.

European Commissioner for the Internal Market Thierry Breton welcomed the result, saying that "Europe is now a global standard-setter in AI."

As the first-ever comprehensive legal framework on AI globally, the regulation has adopted a risk-based approach, subjecting higher-risk AI applications to stricter regulation.

Under the new rules, certain AI applications will be prohibited, including biometric categorization systems based on sensitive characteristics, and the untargeted scraping of facial images from the internet or CCTV footage for facial recognition databases.

Additionally, the Act prohibits emotion recognition in workplaces and schools, social scoring, predictive policing, and AI designed to manipulate human behavior or exploit vulnerabilities.

For general-purpose AI (GPAI) systems, the Act requires compliance with transparency requirements, such as adhering to European Union (EU) copyright laws and publishing detailed summaries of the content used for training.

People visit the Mobile World Congress (MWC) 2024 in Barcelona, Spain, Feb. 28, 2024. (Xinhua/Gao Jing)

Bird and Bird, an international law firm, noted that negotiations surrounding the AI Act brought several contentious issues to light, particularly regarding the treatment of biometric AI systems.

The company highlighted nuances in the text that could introduce uncertainty within the industry, particularly with regard to prohibitions and restrictions on biometric systems, along with exceptions and caveats.

Furthermore, Bird and Bird identified confusion among businesses regarding obligations related to GPAI models, posing challenges in distinguishing between GPAI models and systems, and understanding their separation from high-risk systems.

"Much work lies ahead that goes beyond the AI Act itself," said Dragos Tudorache, co-rapporteur of the Civil Liberties Committee of the European Parliament. "We must now prioritize bringing this law into action."

The AI Act, pending final endorsement, is expected to come into force in May following its publication in the official Journal.

A participant takes a photo during a forum on the theme of "Computing Power Network: Smarter Network for a Smarter World" during the Mobile World Congress 2024 in Barcelona, Spain, Feb. 27, 2024. (Xinhua/Gao Jing)

The rules will become fully applicable within 24 months of their entry into force, with specific timelines for different aspects such as bans on prohibited practices, codes of practice, general-purpose AI rules, and obligations for high-risk systems.