Gemini May Breach Press Publishers’ Rights, Says EU Court Adviser

Google’s Gemini chatbot may infringe the rights of press publishers when it fetches news articles in real time and reuses them in its answers, an adviser to the EU’s top court said on 3 November, in the first clash between Gemini and press publishers to reach the Court of Justice. Advocate General Maciej Szpunar gave his view in Case C-250/25, according to the court’s press release, which calls it “the first generative AI case to reach the Court of Justice”.

This is an opinion, not a ruling. The press release says the opinion “is not binding on the Court of Justice”, that the judges “are now beginning their deliberations”, and that judgment “will be given at a later date”. Reuters noted that the court’s judges usually follow Advocate General opinions.

The case was brought by Like Company, which the press release describes as “a Hungarian operator of various web portals”. It sued Google Ireland, the Google company named in the case, before a regional court in Budapest, the Budapest Környéki Törvényszék. Like Company relies on the right EU copyright law gives press publishers over the online use of their publications. It says Google used its articles to train the large language models behind Gemini, and reproduced and made them available through the chatbot between 13 June 2023 and 31 May 2024.

The Hungarian court paused the case and sent four questions to Luxembourg, in a request lodged on 3 April 2025. Two ask whether training a model on press content is reproduction, and if so whether the text and data mining exception covers it. The other two ask whether a chatbot answer that repeats part of an article counts as reproduction or as communication to the public. One question even asks whether it matters that the chatbot “merely predicts the next word on the basis of observed patterns”.

Gemini and press publishers: the two tests Szpunar proposes

Gemini uses retrieval-augmented generation, which the press release describes as technology that lets the models “collect information in real-time from the World Wide Web”. To answer news questions, the chatbot often retrieves, or “scrapes”, articles online. Szpunar says that when such a chatbot takes a publication from a publisher’s website, reuses it to write an answer and gives that answer to a user, “the press publisher’s right may be infringed in certain circumstances.”

He sets limits first. The right “does not protect the mere facts reported in press publications, but only the specific form of expression”, and is not meant to give publishers a monopoly over the news itself. Within that, he points the court to two factors:

• whether the chatbot accessed the article unlawfully, “in particular by circumventing effective technical measures (such as a paywall) designed to restrict access to, or web-scraping of, that publication”;
• whether the answer merely reports facts, or reuses the article’s form of expression “in a recognisable manner”, to the point that the user “is entirely dispensed from accessing the original publication”.

The making-available right is, as a rule, infringed where the article’s expression is “systematically recognisable” in the answers, he adds. On liability, he suggests the chatbot provider is responsible where the infringing acts are “inherent to the business model of that provider”, or where it has failed to take “appropriate measures in order to counter those acts credibly and effectively”.

Training questions set aside, but with a steer

On AI training, the Advocate General advises the court to declare both questions inadmissible. The press release says there is “no evidence that Like Company’s articles have been used to train the Gemini LLMs”, and that training a model is not a form of “online use” of press publications, so the questions are hypothetical. “For the sake of completeness”, though, he suggests that using protected works to train a language model is, “as a rule, freely permitted under the text and data mining exception.”

The line is drawn at copying expression, not at reporting news

The opinion moves the fight away from training, where most AI copyright suits have centred, and towards what a chatbot shows the reader. That matches a split TSN noted last week, when the Wikimedia Foundation told a New York court that training is likely fair use but that fetching and reproducing content at answer time should follow licences. Szpunar reaches a similar place under EU law, by a different route.

For AI companies, the two tests are concrete enough to act on. Respecting paywalls and anti-scraping measures becomes a legal issue, not just good manners, and answers that closely echo an article’s wording carry more risk than short factual summaries. The liability test also matters: a provider that builds reuse into its business model, or does too little to stop it, could be held responsible. That puts the focus on product design choices rather than on individual answers.

For publishers, the opinion is narrower than a blanket right to be paid. Facts stay free, and a chatbot that reports the news in its own words may not infringe at all. The question will be how recognisable is too recognisable, and that will be for national courts to judge case by case. It also sits alongside voluntary moves, such as the shared AI training opt-out that Google, OpenAI and Microsoft backed on 30 October, and regulatory ones, such as the UK competition watchdog’s investigation into Meta’s terms for publishers. In the US, publishing groups have urged a court to reject OpenAI’s fair-use defence. TSN tracks these cases on its AI copyright lawsuits hub.

What happens next

The judges in Luxembourg now deliberate, and no date has been given for the judgment. Once the court rules, the case goes back to Budapest, where the Hungarian court must decide whether Gemini’s answers about Like Company’s articles crossed either line. Until then, Szpunar’s two tests are the clearest sign yet of how the EU’s top court may treat AI chatbots that read the news for their users.

Sources

• Court of Justice of the EU: Press release No 155/26, Opinion of Advocate General Szpunar in Case C-250/25 Like Company, 3 November 2026
• Court of Justice of the EU: Request for a preliminary ruling from the Budapest Környéki Törvényszék (Hungary), Like Company v Google Ireland Limited (Case C-250/25), lodged 3 April 2025
• Reuters via Investing.com: “Google’s use of press by AI chatbot may infringe copyright rules, EU court adviser says”, 3 November 2026

Related stories

• AI Copyright Lawsuits: Cases, Rulings and Deals
• Wikimedia Brief Says AI Training Is Likely Fair Use, but RAG Is Not
• Google, OpenAI and Microsoft Back a Shared AI Training Opt-Out