First AI-Related Reference to the CJEU: Preliminary Questions in Case C-250/25 Like Company v. Google Ireland

On 3 April 2025, the Budapest Környéki Törvényszék (Hungary) referred a request for a preliminary ruling to the Court of Justice of the European Union (CJEU) pursuant to Article 267 TFEU, in proceedings between Like Company, a Hungarian press publisher, and Google Ireland Limited, a subsidiary of Alphabet Inc. This marks the first time that the CJEU is formally seized of legal questions specifically addressing the functioning and legal qualification of generative artificial intelligence systems, a development likely to herald an increasing wave of AI-related litigation in Europe, in line with the trajectory already observed in the United States.

‍

Background of the Dispute

The applicant, Like Company, operates several online news portals protected under intellectual property law and monetised via advertising revenues. The defendant, Google Ireland, provides access in the European Economic Area to the chatbot known as Gemini (formerly Bard), based on large language models (LLMs), which generates content through probabilistic pattern recognition and user prompts.

The core factual element of the dispute concerns the reproduction and summarisation by Gemini of content from the applicant’s press articles, including an article relating to Hungarian celebrity Kozsó, upon request from users. The applicant contends that the chatbot unlawfully reproduced and made available to the public portions of its protected editorial content, in violation of EU and national copyright rules, between 13 June 2023 and 7 February 2024.

‍

Legal Basis and Questions Referred

The referring Hungarian court seeks guidance on the interpretation of:

• Article 15(1) of Directive (EU) 2019/790 (DSM Directive), relating to the new ancillary rights of press publishers;
• Articles 2 and 3(2) of Directive 2001/29/EC (InfoSoc Directive), concerning reproduction and communication to the public;
• Article 4 of Directive 2019/790, regarding text and data mining exceptions;
  in light of a factual situation involving the generative output of an LLM-based AI system.

The four questions posed by the Hungarian court are as follows:

1. Communication to the Public: Whether the display, by an LLM-based chatbot, of text partially identical to protected press content constitutes an act of communication to the public under Article 15(1) of the DSM Directive and Article 3(2) of the InfoSoc Directive, and whether the predictive mechanism of the AI system alters that legal qualification.
2. Reproduction in Training: Whether the training of a chatbot using LLM techniques—based on the observation and statistical modelling of linguistic patterns—constitutes an act of reproduction under Article 2 of the InfoSoc Directive and Article 15(1) of the DSM Directive.
3. Text and Data Mining Exception: If such training is deemed to involve reproduction, whether it qualifies for the exception for text and data mining under Article 4 of the DSM Directive.
4. Responsibility for User-Prompted Output: Whether the reproduction of protected content by the chatbot in response to a user prompt—which either quotes or refers to the original press publication—can be imputed to the chatbot provider under the same provisions.

‍

Legal Arguments and Positions

The applicant argues that Gemini’s responses amount to acts of unauthorised reproduction and communication to the public, which exceed the scope of any permitted use. It contends that the length and nature of the extracted content surpasses the “very short extracts” permitted without consent under Article 15(1) DSM, and that no valid consent was granted. The applicant further asserts that training the LLM on Hungarian journalistic works constitutes a reproduction outside the limits of lawful text and data mining.

The defendant, by contrast, denies any infringement. It claims that the chatbot does not reproduce or make content available to a “new public,” as the underlying press publications are already accessible online. It further invokes the exceptions for temporary acts of reproduction and lawful text and data mining, and disputes the factual equivalence between chatbot outputs and the original content, citing the phenomenon of AI “hallucination.”

‍

Significance and Outlook

This reference (Case C-250/25) represents a pivotal moment for European copyright law in the age of artificial intelligence. It squarely raises the issue of whether and to what extent generative AI systems fall within the scope of existing copyright directives, in particular the new press publishers’ right under the DSM Directive, and how to reconcile those rights with the legitimate use of AI under the text and data mining exceptions.

The questions submitted by the Hungarian court highlight the need for clarification at the highest judicial level in the EU, as national courts begin to grapple with AI-generated outputs that may intersect with existing intellectual property protections.

In particular, the case opens a path for jurisprudential developments on:

• the scope of “communication to the public” in LLM-generated responses;
• the legal characterisation of AI training practices under reproduction rights;
• the application of new exceptions in the digital environment to cutting-edge technologies;
• the responsibility of AI service providers for the outputs generated by their systems.

It is likely that this preliminary ruling will become a foundational decision in the European legal framework governing artificial intelligence and intellectual property. The outcome will be closely watched by publishers, AI developers, rights holders, and legislators, as it will shape the normative boundaries for AI innovation and copyright enforcement across the EU.