The European Parliament has adopted the law on artificial intelligence (AI) by a large majority. The aim of the new regulation is to promote trustworthy AI in Europe and worldwide. To this end, it will ensure that AI systems respect fundamental rights, safety and ethical principles and also address the risks of high-performance AI models.
EU Commission President Ursula von der Leyen welcomed the clear vote and emphasized that it will strengthen Europe's potential of talents and set a global standard for trustworthy AI. EU Commissioner Thierry Breton praised the law as the result of many years of preparations, consultations and negotiations. Breton emphasized that the law is a balanced, risk-based and future-proof regulation that ensures transparency and promotes the exchange of information between developers and companies.
The regulations will follow a risk-based approach in the future. Accordingly, they will align with the level of assessed risk of an AI application:
- Minimal risk: AI applications with minimal risk (e.g. recommendation systems or spam filters) will receive a free pass and reduced requirements, whereby companies can voluntarily commit to additional codes of conduct.
- High risk: AI applications classified as high-risk will have to meet higher requirements. Such applications include critical infrastructure, such as AI applications in the areas of water, gas and electricity supply, medicine, education and law enforcement, as well as biometric identification, categorization and emotion recognition systems.
- Unacceptable risk: In the future, AI systems that pose a clear threat to fundamental rights will be banned. This includes, for example, applications that manipulate behavior, enable "social scoring" or applications in predictive policing.
- Specific transparency risk: Consumers should be informed more clearly as soon as they come into contact with AI-related applications and AI-generated information. This should be enabled, for example, through clearer labeling and actively provided information.
- Fines: Fines are to be imposed for breaches of regulations.
Other aspects of the law concern the handling of AI applications for general purposes and provisions relating to risk management. An AI office will also be set up in the European Commission to ensure implementation and enforcement and for administrative purposes. The European Commission is also involved internationally in promoting regulations for trustworthy AI.
The regulation will undergo a final review and is expected to be finally adopted before the end of the legislative period. It must be formally approved by the Council. The AI law will come into force 20 days after its publication in the Official Journal and will be fully applicable two years later, with different regulations coming into force at different times. To facilitate the transition, the Commission has launched the AI pact to support implementation and encourage AI developers to meet the obligations of the law ahead of time.
Source: EP
More information