EU Court Adviser: Gemini May Infringe Press Rights

On November 3, 2026, Advocate General Maciej Szpunar of the EU Court of Justice concluded that Gemini's answers may infringe a press publisher's rights when they reuse its text in a recognizable way. It is not a ruling: the Court will decide later. It matters to news sites and to any site publishing original content.

What we know

The case: according to the Court's press release No 155/26, Like Company, which runs news portals in Hungary, sued Google Ireland over the Gemini service. A Budapest court asked the EU Court how EU law applies (Case C-250/25). The official summary of the request shows it was lodged on April 3, 2025.
The questions: four in total. Two about the chatbot's answers and two about training the model.
Training: the Advocate General proposes not answering the two training questions, because there is no evidence the articles were used to train Gemini. He adds that training models is, as a rule, permitted under the text and data mining exception.
Chatbot answers: per the press release, a chatbot that retrieves an article and reuses it to answer a user may, in certain circumstances, infringe the publisher's right.
What is protected: the specific form of expression, not the facts or the news itself. Reuters quotes him saying the rules do not protect the "mere facts."
Two factors: whether the article was accessed unlawfully, for example by circumventing a paywall or another effective technical measure, and whether the answer only reports facts or reuses the wording so closely that the user no longer needs the original.
Liability: the provider would be liable where such uses are inherent to its business model, or where it does not take adequate, credible and effective measures against them.

What changes and what doesn't

Nothing changes yet. The opinion does not bind the Court, which has just begun deliberating. The Hungarian court will then decide the actual dispute.
The debate has a reference point. Reuters notes that judges usually follow the Advocate General's opinions, though they are not required to.
Summarizing news is not made illegal. Under this reading, retelling the facts of a story is not using the protected expression.
It is EU law only. It interprets EU directives. Sites outside the EU are not covered by this reading, although the case may shape negotiations with AI companies elsewhere.

How to tell if it affects you

If your site publishes original text (news, guides, analysis), here is what you can check today:

Test what AI tools return. Ask Gemini and other assistants for a detailed summary of one of your most-read pieces. Does it retell the facts in other words, or repeat your sentences?
Keep dated evidence. If it repeats your text, save the prompt, the full answer, the date and your article's URL. Any claim would start there.
Check what your robots.txt actually controls. Google's documentation says the Google-Extended token governs whether content Google crawls is used to train Gemini models and for grounding, meaning content from the Search index given to the model at answer time. Google says it does not affect inclusion in Search and is not a ranking signal.
Review your access controls. The opinion gives weight to how an article was reached. If you have paid or members-only content, confirm it really sits behind controlled access and is not exposed in the page source.
Write in your own voice. Expression is what is protected. Text with your own analysis, data and style is more recognizable than a rewritten press release.

This article is not legal advice: it summarizes what the Court's press release and Google's documentation say and what that means for a website.

Sources

Updates: when the Court of Justice rules in Case C-250/25, it will be added here with a link.