EU court adviser says Google's Gemini can infringe news copyright

An adviser to the European Union's top court said on November 3 that Google's Gemini chatbot can infringe news publishers' copyright. In fact, it's the first opinion on generative AI to reach the EU's Court of Justice. Read more: EU adviser: Google Gemini can infringe news copyright

Advocate General Maciej Szpunar wrote that the EU's text-and-data-mining exception generally allows training large language models like Gemini on copyrighted text, as long as the text was obtained lawfully. But he said a chatbot's answers can infringe in two situations.

One is when the chatbot's retrieval step gets around a paywall or a scraping block. The other is when its answer reproduces an article's wording closely enough to substitute for the original. Copyright never protects the underlying facts in a news story, only how those facts are expressed.

Szpunar illustrated the distinction with a real example from the case. A Hungarian news site, balatonkornyeke.hu, reported in July 2023 that a singer named Kozsó wanted to bring freshwater dolphins to an aquarium on the Tihany peninsula at Lake Balaton.

When a user asked Gemini to summarize the article, the chatbot added details that were not in the original story. It also invented exact pool dimensions: 25 by 15 by 5 meters, holding 1,000 cubic meters of water. It added public criticism, too, that the dolphins would struggle to adapt and the pool would harm the environment.

Google calls this kind of invented detail a hallucination, a term for when an AI system generates information that sounds specific but isn't backed by its source.

Szpunar called the summary "rather approximative." He said it did not reproduce the article's wording closely enough to infringe, since only the opening sentence appeared in Gemini's answer in anything close to its original form.

Related: Goldman Sachs pulls AI coding tools from traders and bankers

The case against Google

Like Company, which operates balatonkornyeke.hu and other Hungarian portals, brought a copyright case against Google over the Bard and Gemini chatbots in a Budapest court. That court referred four legal questions to the CJEU.

On the dolphin-summary example, Szpunar found it probably did not cross the line into copyright infringement. He noted Like Company used no paywall or anti-scraping measure, and that Gemini's summary didn't track the article's wording closely enough to infringe.

But he said his conclusions "cut both ways," and left open whether Like Company's portals even qualify as press publications under EU law.

The opinion runs 240 numbered points. CJEU judges don't have to follow an advocate general's view, but they usually do, and a final ruling is expected within months, with no date yet set.

A Google spokesperson said the EU's text-and-data-mining exception is vital for European competitiveness and AI development, and that Gemini produces original outputs rather than replicating content.

Related: OpenAI sets shared usage allowance for ChatGPT and Codex

A wider publisher fight

The dispute sits inside a wider fight between online publishers and AI companies. The European Commission opened an antitrust investigation on December 9, 2025, into whether Google uses publishers' content for its AI Overviews feature in Search without compensation or a real opt-out.

Meanwhile, five French publisher groups filed a further complaint with the Commission on October 29, 2026. Reuters reported that Like Company and similar outlets have seen traffic fall as AI-generated summaries increasingly replace search results that carry links back to the original site.

The ruling the CJEU eventually issues will bind courts across the European Union, not just Hungary.