The plaintiff is a Hungarian football fan club. It runs a website on which it regularly publishes articles by various authors about a football club. The referring court has already indicated that this website is a protected press publication under Article 15 DSM Directive, i.e. a collection mainly made up of journalistic literary works.
The defendant offers an AI chatbot in the EU. The plaintiff claims that the defendant used its press publication in the chatbot without consent. In its view, this involved two separate infringements: the defendant reproduced the content, and it made the content available to the public through the chatbot's responses.
The referring court is the Pécs General Court (Pécsi Törvényszék), which hears copyright cases at first instance. It is a different Hungarian court from the one that referred the related case C-250/25.
The decisive legal framework is EU law, so the court has referred six questions to the CJEU. Questions 2, 4, 5 and 6 concern the training of the AI model. Questions 1, 3 and, in part, 6 concern what goes into the chatbot (user prompts) and what comes out of it (responses).
1. AI output: Do AI-generated summaries of press publications infringe the press publishers' right under Article 15 DSM Directive and/or the right of making available to the public under Article 3 InfoSoc Directive?
2. AI training as reproduction: Must Article 15 DSM Directive and Article 2 InfoSoc Directive be interpreted as meaning that processing press publications to operate an AI chatbot constitutes an act of reproduction?
3. Protected content in user prompts: Must Article 15 DSM Directive and Article 2 InfoSoc Directive be interpreted as meaning that an act of reproduction occurs where a chatbot processes a user prompt containing a protected press publication to generate a response?
4. TDM exception: If question 2 is answered in the affirmative, does such reproduction fall under the TDM exception in Article 4 DSM Directive?
5. Implied consent to training and grounding: Must Article 15 DSM Directive and Articles 2 and 3(2) InfoSoc Directive be interpreted as meaning that a press publisher is deemed to have consented to the use of its content for LLM training, tokenisation and grounding if it has not expressly prohibited such use on the service provider's platform?
6. Missing opt-out: Must Article 4(3) DSM Directive be interpreted as meaning that a press publisher is deemed to have authorised the use of its press publications to operate an LLM-based chatbot if it has not expressly and appropriately reserved its rights?