Friday 24 May 2024

Artificial Intelligence Act Receives final approval

The European Union’s Artificial Intelligence Act has received final approval, making it the world’s first major regulation specifically targeting AI. The EU Council formally approved the Act on Tuesday, with the legislation expected to enter into force in mid-June and become effective in 2026. This development is significant for biometrics developers, providers, and users.

Companies that violate the Act may face fines from the EU Commission ranging from 7 million euros (US$7.5 million) to €35 million ($38 million) or between 1.5 and 7 percent of their annual global revenues. SMEs and start-ups will be subject to proportional administrative fines.

The Act's implementation timeline is detailed: General-purpose AI models (GPAI) must adopt obligations 12 months after the Act takes effect, while AI systems in regulated products will have 36 months, according to Reuters.

Since its draft publication in 2021, the AI Act has sparked intense debate. Human rights organizations have criticized it for inadequate protections, while business groups have argued it imposes heavy compliance burdens.

A provisional agreement on the Act was reached in December 2023 following a three-day marathon negotiation, which deviated from the European Parliament's June position advocating a full ban on real-time biometric surveillance. The technical details were finalized in early February 2024, and the Act was passed by the European Parliament in March. The legislation adopts a “risk-based” approach, categorizing AI systems by risk and imposing stricter rules on those deemed harmful to society.

AI applications with unacceptable risk are banned, including those using real-time remote biometric identification, such as facial recognition in public spaces. Exceptions exist for law enforcement in specific cases like kidnappings and terrorism. Other prohibited applications include biometric categorization based on sensitive characteristics, emotion recognition in workplaces and schools, social scoring, predictive policing, and applications that manipulate human behavior. The law also forbids the untargeted scraping of facial images from the web or CCTV footage for creating facial recognition databases.

The legislation introduces transparency requirements, including clear labeling for deepfakes.

To ensure effective implementation, the AI Act establishes new institutions. The AI Office, attached to the European Commission, will coordinate the Act's implementation among Member States and keep classification rules and procedures updated with technological developments. Other bodies include the European Artificial Intelligence Board, the Advisory Forum, and the Scientific Panel of Independent Experts.

Additionally, the law envisions AI regulatory sandboxes for testing AI systems in real-world conditions.



No comments:

Post a Comment