EU court adviser says chatbots reusing news articles may infringe

Advocate General Maciej Szpunar of the Court of Justice of the European Union advised the court on 3 November that an AI chatbot which retrieves press articles in real time and reuses them to answer users “may, in certain circumstances, infringe the rights of press publishers”. The court’s press release called Case C-250/25, Like Company v Google, the first generative-AI case to reach it. The opinion is not binding. The judges have begun deliberating, and judgment will come later.

The dispute started in Hungary. Like Company, which operates web portals, sued Google Ireland in the Budapest Regional Court over its Gemini chatbot (Google is part of the Alphabet group). It argued that Google used its articles to train Gemini’s language models and then reproduced them and made them available through the chatbot between 13 June 2023 and 31 May 2024. The Hungarian court sent four questions to Luxembourg under EU copyright law: two on training, and two on whether chatbot answers fall under publishers’ reproduction and making-available rights. Gemini draws on web content at query time through retrieval-augmented generation, which the press release describes as “grounding”.

On training, Szpunar advised the court to find the two questions inadmissible as hypothetical, because there was no evidence Like Company’s articles had been used for training and because training is not “online use” of a press publication. For completeness he added that using protected works to train a language model is, as a rule, freely permitted under the text-and-data-mining exception in the 2019 copyright directive.

On answers, he said the publisher’s right protects the specific form of expression, not the facts reported. Whether a chatbot infringes the reproduction right depends on two factors: whether it accessed the article unlawfully, for example by circumventing a paywall; and whether the answer reuses the article’s form of expression recognisably enough that the user is dispensed from reading the original. The making-available right is as a rule infringed where that form of expression is systematically recognisable in answers to related queries. The provider is liable, he said, where such acts are inherent to its business model or it has failed to put in place measures to counter them “credibly and effectively”.

The opinion splits the issue in two: training treated generously, retrieval-time reuse treated restrictively. If the court follows it, publishers would have a basis to challenge search-grounded chatbots even where training claims fail, while developers would keep a broad training exception. The sources reviewed contain no response from Google or Like Company. The opinion arrives as publishers press competition and copyright claims over AI use of their content elsewhere.