EU lawmakers reached an agreement on the text of the AI Act on Friday 9 December, following a marathon negotiating session between the European Parliament, European Commission, and Council of Ministers, which represents Member States. The Act, which sets out rules for artificial intelligence across the EU, is based on regulating more harmful uses of AI more strictly, with a lighter touch for less risky cases.
Among the provisions in the final text are the exclusion of free and open-source software from regulation unless their application is considered to be high-risk. Large and higher-risk models will be subject to evaluation and transparency requirements, with a new AI Office to be created in the European Commission to enforce this regulation.
Speaking after the negotiations finished, Dragos Tudorache MEP, who was a co-rapporteur responsible for the Parliament’s position, said, “The EU is the first in the world to set in place robust regulation on AI, guiding its development and evolution in a human-centric direction. The AI Act sets rules for large, powerful AI models, ensuring they do not present systemic risks to the Union and offers strong safeguards for our citizens and our democracies against any abuses of technology by public authorities. It protects our SMEs, strengthens our capacity to innovate and lead in the field of AI, and protects vulnerable sectors of our economy. The European Union has made impressive contributions to the world; the AI Act is another one that will significantly impact our digital future”.
European Commission President, Ursula Von Der Leyen, said, “Today’s agreement focuses regulation on identifiable risks, provides legal certainty and opens the way for innovation in trustworthy AI. By guaranteeing the safety and fundamental rights of people and businesses, the Act will support the human-centric, transparent and responsible development, deployment and take-up of AI in the EU.”
The negotiations took longer to reach agreement on regulating the use of AI in military or defence applications. The final text will include an exemption for this type of use. There was also discussion of which AI practices should be prohibited, with Members of the European Parliament adding in a ban on databases that are based on mass collection of images of peoples’ faces, as well as banning the use of emotion recognition tools in work and education, and predictive policing tools.
Like the EU’s General Data Protection Regulation, penalties for failing to comply with the AI Act will be based on either a minimum amount, or a percentage of a company’s annual global turnover. Fines could reach 7% of turnover or €35 million for the most severe violations.
The Act is expected to take force two years after adoption, with a shorter timescale for provisions about prohibited uses.