Google’s Gemini chatbot could potentially violate press publishers’ rights when it pulls articles from news sites to build its answers, according to an adviser to the European Court of Justice. The new finding could affect all generative AI services operating in the EU.
The opinion, written by Advocate General Maciej Szpunar, draws a line between two ways AI companies use copyrighted news content.
He said AI companies can use protected articles to train large language models. EU law allows this under an exception for text and data mining.
The problem starts after training. Szpunar believes a chatbot could break copyright rules if it takes content from a news article and uses it in a reply to a user. He added that copyright protects the words used in an article, and not necessarily the facts it reports.
First AI case for EU’s top court
Hungarian publisher Like Company sued Google Ireland after Gemini produced summaries of articles from its news portals in reply to user queries. The publisher claims the chatbot reproduced its work without permission and went past the limited use of only short extracts.
The dispute is the very first time the Court of Justice of the European Union has been asked to rule on generative AI and the copyright issues associated. It is docketed as case C-250/25.
Google has made several arguments in its defence. It says Gemini does not store full articles in its memory and that any use of copyrighted content in its replies is allowed under rules covering temporary copies and data mining.
Google also argues that Hungarian law does not apply because Gemini was not trained in Hungary.
Google’s claims and what comes next
A Google spokesperson in their defense of Gemini, claimed the chatbot only learns language patterns to produce original responses rather than copy existing content. The spokesperson added that Gemini gives publishers control over how their content is used, including the option to block its use entirely.
Google also welcomed the opinion’s recognition that EU law allows AI systems to train on publicly available information.
The court does not have to follow an Advocate General’s opinion, but it usually does. The opinion can therefore offer an idea of how the court may rule. A final decision is expected in about six months.
The case is also set to potentially affect publishers across the EU, not just one in Hungary. The court’s final decision will help determine how chatbots across the bloc can use news articles in their replies to users. The EU introduced the press publishers’ right in 2019.
Don’t just read crypto news. Understand it. Subscribe to our newsletter. It’s free .
