Google's Gemini chatbot may violate European copyright law when it reproduces protected news content in its answers, an adviser to the EU's highest court warned on Tuesday, November 3, in an opinion that could oblige technology companies to compensate publishers for certain uses of their journalism.

Advocate General Maciej Szpunar delivered the opinion to the Court of Justice of the European Union (CJEU) in a case brought by Hungarian website operator Like Company against Google, the U.S. technology company behind Gemini. The dispute turns on a single question: whether an AI chatbot infringes European copyright protections when it retrieves published articles and incorporates their content into answers delivered to users.

What did the adviser actually say?

Szpunar set out the circumstances in which publishers' rights may be breached. "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," he said.

Two conditions matter. Infringement may occur when a chatbot accesses protected publications by unlawfully bypassing paywalls or other technical restrictions. It may also arise when AI-generated responses reproduce "the form of expression of that publication in a recognisable manner".

The adviser drew one boundary firmly: copyright protection does not extend to the underlying facts reported in news articles. European rules protect "only the specific form of expression". The distinction matters because AI systems can generally communicate factual information without reproducing the copyrighted language used by journalists and publishers.

What is at stake for publishers?

A ruling against Google could require the company to compensate publishers for certain uses of their material. For digital news organizations, the financial implications are direct: AI chatbots, unlike conventional search engines, deliver summaries within their own interfaces rather than displaying links to original articles, which can reduce traffic to publishers' websites.

The case is part of a wider confrontation. Publishers argue that AI-generated summaries shrink their traffic while allowing platforms to benefit from professionally produced news content without adequate payment. The disputes have already produced complaints before EU and national antitrust regulators.

A favorable ruling could provide revenue support to news outlets losing readers as consumers shift from traditional search results to AI-generated answers.

How does AI training fit in?

Szpunar separated the question of chatbot answers from the question of model training. Large language models are trained on extensive text collections to recognize language patterns, generate responses and summarize information. Using copyrighted material for that purpose is "freely permitted under the text and data mining exception", he said — a legal framework under European law allowing certain automated analyses of copyrighted works.

The adviser also noted there was no evidence that Google had used Like Company's articles to train Gemini.

Google has argued in response that text and data mining exceptions are essential to Europe's technological competitiveness and the continued development of AI. The company says these legal protections support innovation in AI systems.

When will the court rule?

The CJEU, based in Luxembourg, is expected to issue its ruling in the coming months. The advocate general's opinion is not legally binding, but the court follows its advisers' recommendations in a majority of cases.

The eventual decision could clarify when AI-generated responses infringe publishers' rights, and whether technology companies must obtain authorization or pay for certain uses of news content. It may also shape ongoing negotiations among publishers, platforms and regulators over how journalism feeds into AI-powered services.

For U.S. technology companies operating internationally, European court decisions could determine how AI products retrieve, summarize and display copyrighted information across their services.

The ruling will be another significant test of how existing intellectual property law applies to generative AI — and it will set the boundaries for how chatbots use the news.