The role of artificial intelligence in the 21st legislative period: an evaluation of the coalition agreement

At 3pm on 9 April 2025, the party leaders of the CDU, CSU and SPD presented their new coalition agreement for the 21st legislative period. The preliminary political goals of the prospective future coalition for regulating and promoting artificial intelligence (“AI”) had already been reported – you can find our report on the preliminary results here. We will present the most important final negotiation results on the topic of AI below.

A. Basic objective: Germany as an “AI nation”

The new coalition parties have set themselves an ambitious goal: they wish to make Germany an “AI nation” and a “leading hub for future technologies”. AI is to be a key priority on the high-tech agenda and is expected to make a significant contribution to economic growth. Accordingly, tapping the potential of AI plays an integral role in various parts of the coalition agreement.

To achieve the goal of making Germany an AI nation, the future governing parties aim to invest heavily in the cloud and AI infrastructure and in linking AI and robotics. The coalition partners hope that the development of key technologies and the safeguarding of value chains for these industries will reduce digital dependencies. Industrial AI is to be supported by innovation-friendly regulation. To promote Germany’s competitiveness, a commission of experts is to be set up at the Federal Ministry for Economic Affairs and Climate Action (Bundesministerium für Wirtschaft und Klimaschutz – BMWi) to address the issue of “Competition and Artificial Intelligence”.

At least one AI Gigafactory is to be set up in Germany. AI Gigafactories are facilities with supercomputers and data centres that are specifically designed to develop, train and provide next-generation AI models. In addition, an interoperable and European-compatible sovereign Germany stack will be created that integrates AI, cloud services and basic components. At the European level, the Eurostack initiative will be supported.

In addition to this innovation-friendly approach, the parties also address the risks of AI. In particular, key technologies will be developed taking into account the concerns of environmental, social and economic sustainability and ensuring AI security. A possible adjustment of the liability rules for AI is being considered and, if necessary, will be initiated at the European level.

B. Establishing a Federal Ministry for Digitalisation

The future governing parties also plan to establish a new Ministry of Digitisation and State Modernisation. It will probably be headed by Dr Kristina Sinemus, a CDU politician who is currently digital minister of the state of Hesse. Until recently it was unclear whether there would be a federal digital ministry. While such a ministry was proposed in a negotiators’ first paper, a second paper no longer mentioned any such ministry.he scope of the ministry’s responsibilities has not yet been defined in detail. In fact, it is striking that the ministry is only mentioned in the coalition agreement in the context of the distribution of responsibilities between the various ministries. AI is likely to be an important topic for the digital minister in the future. There are already calls for the ministry to be referred to as the AI Ministry.

C. Use of AI in the public sector

The future governing parties have also set ambitious goals for the public administration, planning nothing less than an “administrative revolution”. The vision is for a “networked, efficient and effective” administration that must also be “accessible to all in a user-friendly and low-threshold manner”. To achieve this, automation and AI are to be used explicitly to accelerate and increase the efficiency of administrative processes. The German Administration Cloud is to play a key role in this. The Cloud is planned to encompass a wide range of cloud services and infrastructure for and by the German administration.

The promotion and use of AI is repeatedly emphasised in the various policy areas of the coalition agreement. In the security sector, security authorities are to be given the authority to carry out automated, AI-based data analysis in certain cases. In compliance with constitutional standards, automated data research and analysis and also a subsequent biometric comparison with publicly accessible internet data is to be carried out using AI.

In the financial administration, the use of AI is intended to increase efficiency. The use of AI is also to be made possible in the judiciary. However, it remains unclear which specific applications the coalition partners have in mind for AI in the judiciary. In the healthcare sector, the governing parties advocate AI-supported treatment and care documentation. A digital portal for family services is to include AI features. Even in the cultural sector, a “Culture & AI” strategy is to be developed to exploit the potential of AI here as well, while at the same time mitigating risks. The coalition agreement also addresses the defence sector. According to the agreement, the German armed forces are to be increasingly equipped with future technologies including AI.

D. Implementing the European AI Act

The implementation of the European AI Act (Regulation (EU) 2024/1689), which came into force on 1 August 2024, is mandatory. The AI Act creates a binding European framework for the marketing, commissioning and use of AI for the first time. National legislators must adopt a corresponding implementing law by 2 August 2025. At the end of the last legislative period, the Federal Ministry for Economic Affairs and Climate Action and the Federal Ministry of Justice had already submitted an AI Market Surveillance Bill (KI-Marktüberwachungsgesetz). The coalition agreement does not provide any immediate details of the implementation. However, it states that further burdens on the economy are to be reduced as part of the technical and legal specifications of the AI Act and that an “innovation-friendly and low-bureaucracy” implementation is thus the aim. This is in line with the general maxim of the coalition agreement, which is initially independent of AI, not to carry out what is referred to as “gold-plating”, i.e. bureaucratic over-fulfilment of EU rules when implementing them at national level. However, the controversial demand that was still included in the working group 3 (“digital matters”) summary of results to pursue an adaptation of the AI Act at the European level as a whole is not mentioned in the coalition agreement.

In addition, the AI Market Surveillance Bill will have to stipulate which authority or authorities are to monitor the application of the AI Act (see the first sentence of Article 70(1) of the AI Act). In the past, the data protection authorities have repeatedly offered to act as supervisory authorities, arguing that they are particularly close to the matter and are already monitoring the market in the area of data protection law. This would give market surveillance a federal structure, which, from the point of view of companies, carries the risk of divergent legal interpretation and implementation. According to the AI Market Surveillance Bill the authorities that are already the responsible market surveillance and notifying authorities in fully harmonised areas of product regulation are also the competent authorities in the area of the AI Act. In areas without any such responsibilities, the Federal Network Agency (Bundesnetzagentur) is to become the market surveillance and notifying authority.

While the coalition agreement does not explicitly clarify this question, the future coalition partners want to ensure that market supervision of the AI Act is not fragmented. This means that everything now argues in favour of bundling market supervision at the Federal Network Agency, as already provided for in the AI Market Surveillance Bill.

According to Article 57 of the AI Act, the Member States are obliged to establish at least one AI regulatory sandbox at national level. AI regulatory sandboxes provide a controlled environment to promote innovation and facilitate the development, training, testing and validation of innovative AI systems for a limited period of time prior to their marketing and deployment. The coalition parties are becoming more courageous on this issue. Small and medium-sized companies and start-ups in particular are to be supported by AI regulatory sandboxes. At other places in the coalition agreement the potential of regulatory sandboxes for Germany’s innovative strength is also emphasised and therefore establishing such sandboxes is to be actively promoted. It is likely that Germany will not stop at just one AI regulatory sandbox when implementing the AI Act.

E. Outlook

The coalition agreement addresses the now central importance of AI across a wide range of areas, from the economy to security and administration. In this respect, the future German government is particularly committed to being open to innovation and wants to make Germany an AI nation by providing new impetus. It remains to be seen how this ambitious goal will be implemented in practice. The coalition agreement naturally does not specify any concrete steps. The first interesting indication will be the design of the Federal Ministry for Digitalisation and State Modernisation. In addition, the coalition partners will have to deal with implementing the AI Act as soon as possible. This will show for the first time whether the prospective new government is serious about its proclaimed openness to innovation and cautious approach to regulation.

We would like to thank Cathrine Crämer for her contribution to this publication.