CJEU to rule on AI and copyright in a landmark case against Google

The complex interplay between artificial intelligence and copyright law is set for a defining moment in Europe.

For the first time, the Court of Justice of the European Union (CJEU) has been asked to assess how copyright rules would apply to generative AI technologies, a crucial step in shaping the future of innovation and content protection in Europe.

Context

The applicant, Like Company, is a Hungarian press publisher managing several online news portals offering copyrighted content. It claims that Google’s chatbot, Gemini (formerly Bard), which operates using large language models (LLMs), reproduces and summarises substantial portions of its copyrighted articles without prior authorisation. According to Like Company, Gemini generated responses that echoed significant parts of its original contents, thereby infringing its exclusive rights.

On 3 April 2025, the Budapest Környéki Törvényszék (Budapest District Court) referred Like Company v. Google Ireland (Case C-250/25) to the CJEU, raising the first preliminary questions on generative AI and copyright.

The core issue is whether the training and operation of a large language model (LLM), and the outputs it generates, constitute acts of reproduction or communication to the public under EU copyright law.

The court seeks interpretation of:

Article 2 and Article 3(2) of Directive 2001/29/EC (InfoSoc Directive) on reproduction and communication to the public;
Article 4 and Article 15(1) of Directive (EU) 2019/790 (DSM Directive) on the text and data mining exception and press publishers’ rights.

Questions referred

The CJEU has been requested to respond to four major questions:

Communication to the Public: Does the display, in the responses of an LLM-based chatbot, of a text partially identical to press articles that exceeds the “use of individual words or very short extracts,” constitute an act of communication to the public under Article 15(1) of the DSM Directive and Article 3(2) of the InfoSoc Directive?
Reproduction Right: Does the training of an LLM-based chatbot, by learning and modelling linguistic patterns from protected works, constitute an act of reproduction under Article 2 of the InfoSoc Directive and Article 15(1) of the DSM Directive?
Scope of the Text and Data Mining Exception: If such training is deemed reproduction, is it covered by the Text and Data Mining exception provided in Article 4 of the DSM Directive?
Liability for Outputs: Can the reproduction of protected press content by a chatbot in response to a user prompt, which either quotes or refers to the original publication, be attributed to the chatbot provider under Article 2 of the InfoSoc Directive and Article 15(1) of the DSM Directive.

Arguments of the parties

Like Company argues that Gemini’s outputs constitute unauthorised reproduction and communication to the public under EU law. It contends that the chatbot’s responses exceed the limit of “very short extracts” set by Article 15(1) of the DSM Directive, harming its ad-based revenue by diverting users. The publisher further claims that its consent for indexing its content on search engines does not cover the reuse of that content by Gemini in chatbot responses, which requires separate authorisation. It also argues that the training of Gemini on protected works falls outside the scope of the text and data mining exception and that allowing broader use without consent would undermine the economic protections intended by Article 15.

Google Ireland denies any infringement. It argues that Gemini’s outputs do not constitute communication to a “new public,” since the original articles were already accessible online. It asserts that Gemini generates responses dynamically through probabilistic modelling based on processed training data, without storing or retrieving copies of the original content. Google contends that any resemblance to Like Company’s works is incidental or results from generative AI hallucinations. It further invokes EU exceptions for temporary acts of reproduction and text and data mining and characterises Gemini as a system that produces new and modified outputs rather than reproducing existing content.

Implications

The Like Company v. Google Ireland referral marks a pivotal moment for European copyright law in the era of artificial intelligence. Under the DSM Directive and the Information Society Directive, press publishers benefit from exclusive rights to authorise or prohibit the reproduction and communication to the public of their press publications by information society service providers. The CJEU is now called upon to determine whether, and under what conditions, these rights extend to the use of such publications by generative AI systems like Gemini.

The case raises fundamental issues including whether AI-generated outputs qualify as communication to the public, whether training LLMs on protected works constitutes reproduction, and whether the text and data mining exception can shield commercial AI training. It also tests the allocation of liability for outputs generated in response to user prompts.

A ruling in favour of Like Company could require AI developers to obtain licences for both training and deployment, increasing legal and financial pressures across the AI industry and reinforcing the rights of publishers and content creators. Conversely, a ruling in favour of Google may confirm that AI-generated outputs are sufficiently transformative or indirect to fall outside reproduction and communication rules, enabling broader reliance on publicly available data without compensation.

Last but not least, the Court’s decision will clarify the scope of the Text and Data Mining exception under Article 4 of the DSM Directive, a key issue for large-scale AI training practices.

The highly anticipated ruling of the Court is unlikely to be delivered before 2027.