EU Court to Address AI Chatbots and Copyright in Google Case

A landmark case regarding copyright and generative AI has been presented to the Court of Justice of the European Union (CJEU). This case will explore how European copyright laws pertain to AI-generated content and the training of large language models (LLMs).

The dispute, involving Like Company v. Google Ireland C-250/25, centres on the use of copyrighted press content by Google’s AI chatbot, Gemini. The decision could potentially redefine the responsibilities of AI providers within the EU and clarify the rights of press publishers in today’s digital landscape.

Background

The plaintiff, Like Company, a Hungarian news publisher, claims that Google’s AI service, Gemini, reproduced and distributed parts of its copyrighted articles without approval. The case in question involves content about a Hungarian singer’s initiative to introduce dolphins to Lake Balaton, which Gemini allegedly summarised in detail, including significant portions of the protected material.

The Budapest Környéki Törvényszék, the referring court, has requested the CJEU to interpret key provisions of the 2001 Copyright and Information Society Directive (InfoSoc Directive) and the 2019 Copyright in the Digital Single Market Directive (CDSM Directive). The primary legal issue is whether the operation and output of an LLM can be considered acts of reproduction or communication to the public under EU copyright laws.

Legal Questions Before the CJEU

Does displaying content in chatbot responses that mirrors press articles, beyond a “very short extract,” constitute communication to the public under Article 15(1) of the CDSM Directive and Article 3(2) of the InfoSoc Directive?
Is the act of training an LLM, including tokenisation and learning linguistic patterns from protected works, a form of reproduction under Article 2 of the InfoSoc Directive?
If such training is deemed reproduction, does it qualify for the text and data mining (TDM) exception under Article 4 of the CDSM Directive?
Does generating a response by the chatbot that includes copyrighted content require authorisation from the rightsholder, qualifying as reproduction by the AI provider?

Arguments Presented by the Parties

Like Company argues that both the training and outputs of Gemini unlawfully use protected works. They assert that the chatbot’s responses exceed the permissible limits under Article 15 CDSM, which allows only very short extracts. The publisher claims that their content was systematically reproduced during training, falling outside the TDM exception, which is intended for scientific research and not causing economic harm. They also argue that Gemini’s outputs deter traffic to their website, affecting their ad-based revenue.

Conversely, Google Ireland contends that Gemini does not store or retrieve article copies; rather, it tokenises training data to create new text through probabilistic methods. Google argues that any resemblance to Like Company’s content is coincidental or due to “hallucination,” a known AI phenomenon. The company claims no “new public” is reached since users could access the original content online and cites EU exceptions for temporary reproductions and the TDM exception, positioning Gemini as a creative tool rather than a database.

What We Expect the CJEU to Clarify

This case offers the CJEU a chance to address several key legal points:

Whether AI-generated content that closely resembles copyrighted work constitutes communication to the public under EU law.
Whether converting protected works into tokens for pattern analysis during training is reproduction under Article 2 of the InfoSoc Directive.
Whether the training of LLMs using publicly available content qualifies for the research-related TDM exception, or if commercial use voids this protection.
Whether generating responses that include identifiable elements of protected material constitutes infringement.

Implications for the AI Sector and Publishers

This case marks the first CJEU referral specifically addressing the use of press publisher content by generative AI systems. The outcome could have significant implications for AI developers, publishers, online platforms, and rights holders.

If the CJEU rules that copyright laws apply to both the training and output of LLMs, AI companies may need to secure licences before using or reproducing protected content, increasing legal and financial pressures on AI developers and strengthening the rights of creators and publishers.

Alternatively, if the CJEU determines that AI outputs are too indirect or transformative to be considered reproduction or communication to the public, developers may continue using publicly available data without needing individual licences, potentially encouraging further LLM advancements across Europe, particularly in the commercial sector.

The ruling could also influence the interpretation of the TDM exception under Article 4 of the CDSM Directive, which many AI developers currently rely on for large-scale content ingestion. The CJEU is expected to clarify the limits and applicability of this exception to commercial LLM training.

Copyright Compliance in the Age of AI

Companies in AI development, digital publishing, content licensing, or any data-driven industry should closely monitor this case. Depending on the judgment, organisations might need to:

Conduct due diligence reviews of training datasets and audit for press content inclusion.
Explore contractual approaches to licensing third-party content or excluding high-risk categories.
Review how AI tools generate summaries, quotes, or outputs that may reflect protected works.
Keep abreast of EU and national enforcement trends regarding Article 15 CDSM and TDM exceptions.
Prepare for potential renegotiation or establishment of licences with press publishers, if necessary.

Conclusion

Like Company v. Google Ireland is a pivotal copyright case for generative AI in the EU. The CJEU’s decision will likely provide guidance on the treatment of AI-generated content under the CDSM Directive and whether current LLM training practices fall within existing exceptions. This ruling could serve as a foundational decision with lasting implications for AI development, regulation, and commercialisation in Europe.

For organisations involved in AI development or deployment, this case highlights that regulatory and rights frameworks are evolving. Strategic preparation for any outcome is now essential.