An EU Court Adviser Says Google's Gemini Can Break Copyright by Reusing News Stories, but AI Training Is Fine
An adviser to the EU's top court said Google's Gemini may infringe press publishers' rights when it gets around paywalls or reuses a story's recognisable wording, while training AI models on copyrighted material is allowed.

Europe's highest court has its first serious signal on whether AI chatbots owe anything to the newsrooms they read. On Tuesday, an adviser to the Court of Justice of the European Union said Google's Gemini can infringe press publishers' rights when it lifts their stories to answer users. In the same breath, he said training AI models on copyrighted work is perfectly legal.
Both sides got something. Neither got everything.
A Hungarian publisher against Google
The case was brought by Like Company, a Hungarian website operator, against Google over its Gemini chatbot. Advocate General Maciej Szpunar set out his view in an opinion delivered in Luxembourg.
"When an LLM (large language model)-driven chatbot retrieves a publication from a press publisher's website, reuses that publication to generate an answer, and provides that answer to the user, the press publisher's right may be infringed in certain circumstances," the court's statement said.
Those circumstances are specific. Gemini's answers fall under publishers' right of reproduction when the chatbot gets in by unlawfully circumventing paywalls or other technical barriers. They also fall under it when Gemini reuses "the form of expression of that publication in a recognisable manner."
Facts are free, wording is not
Szpunar drew a clear line around what publishers actually own. Copyright does not protect the "mere facts" a story reports, he said, "only the specific form of expression." A chatbot telling users what happened is not the problem. A chatbot repeating how a journalist wrote it, or sneaking past a paywall to read it, could be.
The AI labs got their own clear sentence. Using copyrighted material to train large language models is "freely permitted under the text and data mining exception," Szpunar said. That takes the biggest threat off the table for model makers, at least in this case: the fight is over what chatbots serve up, not what they learned from.
Why publishers care so much
For outlets like Like Company, the stakes are about survival. Traffic to news sites has been dwindling as AI chatbot summaries increasingly replace the old search results full of links. A ruling that forces Google to pay fees could be a lifeline.
Regulators elsewhere are leaning the same way. A day earlier, Turkey's competition authority ordered Google to offer news publishers fair terms in AI Overviews and to let them opt out without disappearing from search.
Szpunar's opinion is not binding. But the court's judges usually follow their advocates general, and no date has been given for the final ruling. Until then, Europe's publishers have something they did not have last week: a senior legal voice saying a chatbot's answer can be their property.
Some offers on this page may be paid placements or contain affiliate links.
Subscribe to Techpresso
Free daily newsletter, read in 5 minutes.
Subscribe free