The European Parliament has granted final approval to a comprehensive set of regulations aimed at governing the use of artificial intelligence (AI).
The approved legislation, known as the Artificial Intelligence Act, received the green light from lawmakers during a session on Wednesday. European Union officials emphasize that these rules, initially proposed in 2021, are designed to mitigate potential risks associated with the rapid advancement of AI technology while also fostering innovation.
The swift passage of the law by Brussels follows the emergence of ChatGPT, developed by Microsoft-backed OpenAI, in late 2022, which sparked a global race in AI development.
The proposal encountered opposition from only 46 members of the European Parliament in Strasbourg, with a significant majority of 523 MEPs lending their support.
Formal endorsement of the legislation by the European Council is anticipated by May. The rules will become fully enforceable 24 months after they come into effect.
Key provisions of the regulations encompass the oversight of high-impact general-purpose AI models and high-risk AI systems, mandating compliance with specific transparency requirements and EU copyright laws.
The act extends its regulatory scope to cover foundational models or generative AI, such as OpenAI, which are trained on vast datasets to generate new content and execute tasks.
Government utilization of real-time biometric surveillance in public spaces will be limited to specific circumstances, including the prevention of serious crimes, the interception of credible threats like terrorist attacks, and the pursuit of individuals suspected of major offenses.
Brando Benifei, an Italian lawmaker who spearheaded the legislation alongside Romanian MEP Dragos Tudorache, hailed the occasion as a historic milestone on the journey toward AI regulation.
“This is the first regulation in the world that is laying out a clear path toward the safe and human-centric development of AI,” stated Benifei. Tudorache echoed this sentiment, emphasizing the delicate balance achieved between fostering innovation and safeguarding interests.
Thierry Breton, the EU’s internal market commissioner, celebrated the parliamentary endorsement, asserting that Europe has now established itself as a global leader in promoting trustworthy AI.
Under the EU’s risk-based approach, more stringent requirements are imposed on systems deemed to pose greater risks, with outright prohibitions on AI tools considered to be particularly threatening.
For instance, providers of high-risk AI must conduct thorough risk assessments and ensure compliance with legal standards before their products can be introduced to the market.
Breaches of these regulations could result in substantial fines ranging from €7.5 million to €35 million ($8.2 million to $38.2 million), depending on the nature of the violation and the size of the company.
The regulations also impose strict bans on the use of AI for predictive policing and systems that employ biometric data to infer sensitive personal attributes such as race, religion, or sexual orientation.
Moreover, real-time facial recognition in public spaces is prohibited in most instances, with exemptions granted to law enforcement agencies subject to prior authorization from judicial authorities.