EU approves landmark AI Act that was proposed in 2021
European Union approves Artificial Intelligence Act, aiming to ensure safety, compliance with fundamental rights and foster innovation, prevent complications caused by AI
In a historic move, the European Parliament Wednesday approved the far-reaching Artificial Intelligence Act, setting the stage for a regulated and human-centric development of AI in the European Union (EU).
The legislation, first proposed in 2021, aims to ensure safety and compliance with fundamental rights and foster innovation while addressing the potential risks associated with the rapid evolution of technology.
The legislation agreed upon in negotiations with member states in December 2023, received overwhelming support in the European Parliament with 523 votes in favor, 46 against and 49 abstentions.
The EU's 27 member states are expected to endorse the text in April, with the law scheduled for publication in the EU's Official Journal in May or June.
"Europe now a global standard-setter in trustworthy AI"
EU officials emphasize that the rules, prompted in part by the emergence of OpenAI's Microsoft-backed ChatGPT in late 2022, are designed to safeguard citizens in the face of a rapidly advancing AI landscape.
ChatGPT's impressive capabilities, from processing complex text to generating poems and passing medical exams, accelerated the need for comprehensive regulations.
Italian lawmaker Brando Benifei and Romanian MEP Dragos Tudorache led the effort to push the legislation through parliament. Tudorache highlighted the delicate balance achieved between fostering innovation and protecting the interests of the public.
Banned applications
The AI Act adopts a risk-based approach, imposing stricter requirements on riskier AI systems and outright bans on tools deemed most threatening. Prohibited applications include biometric categorization systems based on sensitive characteristics, untargeted scraping of facial images for recognition databases, emotion recognition in workplaces and schools, social scoring, predictive policing solely based on profiling and AI manipulating human behavior.
The legislation also addresses the use of biometric identification systems by law enforcement, imposing strict safeguards on their real-time deployment.
Violations can result in fines ranging from 7.5 million to 35 million euros ($8.2 million to $38.2 million), depending on the infringement type and company size.
Despite lobbying pressures, the EU resisted calls to exclude large AI models from the regulation. Concerns about potential weakening through corporate lobbying remain, with watchdogs pointing to efforts by startups like Mistral AI and Aleph Alpha, along with tech giants Google and Microsoft.
Officials emphasized the regulation's balanced, risk-based, and future-proof nature, standing firm against special interests and lobbyists during negotiations. However, watchdogs caution that further clarification is needed on various aspects of the AI Act, including standards, thresholds and transparency obligations.
Source: Newsroom
