Artificial intelligence poses two key challenges for the media industry: On the one hand, copyright-protected content is used by AI systems without remuneration. On the other hand, AI providers are undermining the business models of many media companies by offering AI-generated summaries, meaning that the media companies' own articles are read less or no longer at all – this also jeopardizes the basis of free media. This creates a double loss situation for publishers and broadcasters. At the same time, high-quality media data is essential for the development of competitive and trustworthy AI models.

Host Carsten Schwecke, Chief Commercial Officer and responsible for all commercial, technological, and data- and AI-based activities at RTL Germany, emphasized in his keynote speech: "Europe needs a strong, sovereign, and independent media ecosystem—especially in times when artificial intelligence is rewriting the rules of digital competition. When international platforms use our journalistic content for their AI models without consent or compensation, it's not just about economic interests, but about the foundation of democratic public discourse. Germany and France together have the expertise, responsibility, and cultural mission to create an alternative to this imbalance: through clear rules for copyright and related rights, trustworthy European AI models, and innovation-friendly regulation. Our goal is to protect and promote editorial content, ensure diversity of opinion, and at the same time develop innovation—so that European media companies can remain independent in the AI era, ensure diversity, and fulfill their central role in an informed, democratic society."

Stefan Heijdendael, advisor public affairs at NDP Nieuwsmedia, presented a Dutch AI model as a best practice example, which was trained using licensed press data and integrated into a fair remuneration system. While Heijdendael emphasized the advantage of closed-source models in protecting the interests of publishers, Dr. Nicolas Flores-Herr, Team Lead Foundation Models and GenAI Systems at Fraunhofer IAIS, presented current European projects for training AI Models that aim to fuel innovation and strengthen collaboration using an open-source approach. The following breakout sessions discussed the most important organizational, political, and technical challenges as well as possible solutions:
 

Protection of business models, establishment of a fair remuneration system

The participating publishers and broadcasters are calling for the protection of their business models and media content – for example, through technical measures such as the effective blocking of data crawlers and the introduction of monitoring systems to prevent the unauthorized use of content and to be able to detect violations. They also want a licensing system that compensates them when AI models are trained with their data. The European media ecosystem has so far lacked a common understanding of how media content should be remunerated.
 

Digital sovereignty and innovation-friendly regulation

Participants would like to see stronger support from politicians at both the national and European level, for example through the use and promotion of EU-compliant AI models. In addition, the development of European alternatives – for example, in the form of AI platforms with diverse media content and links to original sources – was discussed. AI-generated summaries should also be regulated and the legal situation surrounding text and data mining (TDM) should be clarified quickly. Another key concern was transparency in the training of AI models: It must be clear which data sets are being used, and opt-out rules must be respected.
 

Access to infrastructure and high-quality data

The development of proprietary European large language models (LLMs) and foundation models specifically for the media sector, that are built on the basis of data from publishers and broadcasters, was also discussed. In order to remain competitive internationally, greater cooperation between existing European tech teams would be helpful. This would require the availability of high-quality data, curated by humans wherever possible, as well as access to suitable cloud infrastructure and the integration of user experiences. In addition, economic issues—such as the development of sustainable business models and the monetary valuation of content—should also be incorporated more strongly into research approaches in the future.

Finally, ensuring media diversity, democracy, and freedom of the press through AI solutions that promote pluralism and diversity of opinion was identified as a core social task – a unified approach by the industry and intensive stakeholder dialogue are essential for this. Dr. Joachim Köhler, department head at Fraunhofer IAIS, summarizes: "The European media and technology industry faces the challenge of developing innovative AI solutions that meet the interests of rights holders as well as the requirements for transparency and quality. We want to explore ways to create a balanced and innovation-friendly ecosystem for foundation models in Europe – through close cooperation between research, industry, and media companies."
 

Outlook: Presentation of results and next steps

The "French-German Media Industry AI Day" took place following the first "French-German AI Industry Executives' Dialogue" organized by the Fraunhofer ICT Group and INRIA with support from Institut Mines-Télécom (IMT). The aim is to establish partnerships and collaborations in various sectors and to promote a joint AI initiative in Europe. The results of the seven industry workshops will be presented at the Adopt AI International Summit in Paris on November 25-26, 2025. Follow-up workshops and further presentations of results are already being planned, including at the "2nd AI Executive Dialogue" in Paris in 2026.
 

Statements