The European Union has taken a pioneering step by adopting the world’s first Artificial Intelligence Act, setting a potential global benchmark for future AI regulations. Announced by the EU Council, the legislation aims to foster the development and deployment of safe and reliable AI systems within the EU’s single market, while ensuring the protection of fundamental rights of its citizens and promoting investment and innovation in AI across Europe.
Belgian State Secretary for Digitalization, Mathieu Michel, hailed the act as a milestone, addressing global technological challenges that present opportunities for society and the economy. The law, which will be published in the EU Official Journal in the coming days and become effective 20 days post-publication, applies only within the realms of EU legislation, with specific exemptions for military and research purposes.
The act categorizes AI systems according to the risk they pose, with stringent regulations for high-risk applications. It imposes transparency requirements on low-risk applications like video games or spam filters, while high-risk AI systems must meet a series of stringent criteria to gain EU market approval.
Significantly, the legislation bans AI practices that socially profile and categorize individuals based on characteristics such as gender, race, political beliefs, or sexual orientation. It also addresses AI systems considered to be of unacceptable risk, including those used for indiscriminate surveillance and predictive policing based solely on profiling.
This legislative move comes as AI usage has nearly doubled in the last six months, with three-quarters of employees reportedly using AI tools at work, according to studies by Microsoft and LinkedIn. The act also ensures that citizens can lodge complaints against AI systems that infringe upon their rights, marking a critical step in regulating the rapidly evolving technology.