EU Framework for artificial intelligence in place: AI Regulation gains unanimous approval

On 2 February 2024, the EU Member States unanimously approved the Regulation laying down harmonised rules on artificial intelligence (AI Regulation, AI Act) through the Permanent Representatives Committee (Coreper). The AI Regulation forms the EU framework for the use of artificial intelligence (AI) in Europe. Its objectives are to promote innovation, to build trust in AI and to ensure that the technology is used in a way that is compatible with fundamental rights and with the security of EU citizens. The AI Regulation is the world’s first comprehensive AI framework.

Says Dr. Robert Habeck, Vice-Chancellor and Federal Minister for Economic Affairs and Climate Action: “The approval by the EU Member States has brought the long and intensive negotiations on this to a successful end. The AI Regulation is designed to ensure that we in Europe can harness the enormous potential of AI without losing sight of the risks. This balancing act has succeeded. We have been highly engaged in the negotiations to make this happen. As we implement the rules, we will focus on being pro innovation, on creating legal certainty for companies, and on keeping structures lean and free from red tape. We want Europe to be a strong centre for AI.”

Federal Minister of Justice Marco Buschmann commented: “Today is a good day for innovation and for the fundamental rights in Europe. This is because, today, the Permanent Representatives Committee adopted the AI Regulation. This decision paves the way for a secure framework for artificial innovation that promotes innovation whilst addressing risks in its application in an appropriate way. The Regulation strikes the right balance between innovation and protection from risks. This legal framework is the first of its kind internationally. Europe has become a pioneer in this field. That said, it is also clear that we must monitor and evaluate the developments taking place in the field of AI. This is the only way we will be able to continue to ensure the right balance between innovation and protection from risks.”

The Federal Ministry for Economic Affairs and Climate Action and the Federal Ministry of Justice have shared lead responsibility for representing Germany’s position in the European negotiations on the AI Regulation. The following milestones have been reached:
It has been made clear that the AI Regulation is to regulate products, not research and development activities. At the same time, the Regulation forms the basis for the development of user-centred AI in Europe.

There will be no AI-driven social scoring or emotion recognition in the workplace in Europe. The Regulation sets out strict and limiting requirements for the use of remote biometric identification and prevents the use of nationwide biometrical surveillance.

These restrictions are characteristic of the risk-based approach on which the AI Regulation is based: the greater the risk, the stricter the obligations. The use of AI systems deemed to be associated with inacceptable risks (such as social scoring) is prohibited all together, high-risk AI systems are subject to strict technical and organisational requirements, and low-risk applications are merely subject to certain transparency and information obligations.

There will be special rules governing generative AI, especially general-purpose AI models, including ones that generate text and images. If an AI model is particularly powerful and represents a systemic risk, it will be subject to stricter conditions.
More specifically, there will be codes of practice drawn up in cooperation with suppliers and stakeholders until the time when European harmonised standards are available for this purpose. This approach was chosen due to the highly dynamic development of AI and is to ensure that users can access the information they need. Not least due to the advocacy of the Federal Government, the obligations are limited and designed to be as straightforward as possible.

Transparency will also make it easier for small and medium-sized enterprises (SMEs) to use AI. Overall, the needs and interests of SMEs and startups have been given strong consideration – for instance in the rules on regulatory sandboxes that were proposed by the Federal Government and that will create freedom for testing innovations.

The European Parliament and one Council configuration need to formally approve the AI Regulation before it can enter into force. This will then happen on the 20th day after publication in the EU Official Journal. Application will, in principle, begin 24 months after that. However, some provisions will be applicable before then: the bans will already apply after six months, the rules governing general-purpose AI models after twelve months.

As implementation begins, the Federal Government will work with other Member States and the European Commission to advocate a solution that is pro-innovation and does not create red tape.