Policy

EU Court Adviser Backs Publishers in Google Gemini Copyright Fight

An adviser to the EU's highest court has said that AI chatbots could infringe publishers' rights when they reproduce protected works in their answers — but that Google may still escape liability. Advocate General Maciej Szpunar issued the non-binding opinion on November 3, 2026, in a closely watched copyright case brought by Hungarian website operator Like Company. The Court of Justice of the European Union is expected to rule in the coming months, and if it adopts the opinion, the decision could reshape how publishers claim against AI companies — and how those companies build their answers.

What the opinion says

The core of Szpunar's opinion is a distinction between two different things an AI system does: training a model, and reusing protected expression in chatbot answers. Training — the process of feeding text into a model so it learns patterns — is one activity. Generating an answer that reproduces a publisher's protected work is another. The opinion treats the second as the more legally exposed of the two: a chatbot that reuses protected expression in its output could infringe the publisher's rights, even if the training that produced the model was lawful.

The second half of the opinion is the qualification. Even if a chatbot's output could infringe in principle, Google may escape liability in this particular case. The opinion alone does not require Google to pay fees, and the details of why Google may escape — the specific legal mechanism — will matter enormously when the court rules. An opinion that finds infringement in principle but no liability in practice is a very different outcome from one that finds liability outright.

The case behind it

The case was brought by Like Company, a Hungarian website operator, and it concerns Google's Gemini chatbot. Like Company's complaint is, in essence, that Gemini's answers draw on its published material in a way that infringes its rights as a publisher. The case reached the Court of Justice of the European Union because the legal questions it raises — about how EU copyright law applies to AI-generated answers — are questions that no national court can settle alone.

The case is closely watched because it is one of the first to put the specific question of chatbot output before the EU's highest court. There have been lawsuits about training data, about scraping, and about the use of copyrighted material to build models. This case asks a narrower and more consequential question: not whether the model was trained on your work, but whether the answer it gives your reader is your work, reproduced.

Training versus reuse: why the distinction matters

The distinction Szpunar draws is the most important part of the opinion for anyone who publishes on the web. If the law treats training and reuse as separate activities, then an AI company could win on training and lose on output — or the reverse. A model trained lawfully could still produce infringing answers, and a model trained unlawfully might produce answers that are perfectly lawful. That separation complicates every argument in the debate, because both sides have been treating 'AI used my content' as a single claim.

For publishers, the distinction is useful: it gives them a claim that does not depend on proving what happened inside a training run, which is usually invisible from the outside. An answer is visible. If a chatbot's answer reproduces your article, you can point at it. For AI companies, the distinction is a warning: even a clean training pipeline does not guarantee clean outputs, and the output is where users — and courts — will look.

What happens next

The opinion is non-binding. Advocate generals advise the court; they do not decide cases. The Court of Justice of the European Union is expected to rule in the coming months, and it may adopt the opinion, reject it, or adopt part of it and not the rest. Until the ruling arrives, the opinion is a strong signal of the court's likely direction, but it is not the law.

If the court does adopt the opinion, the consequences will ripple well beyond this case. Publishers across the EU would have a clearer basis for claims against chatbot operators whose answers reproduce their work, and tech companies would need to think harder about how their systems generate answers — not just how they train models. The opinion alone does not require Google to pay fees, but a court ruling that follows it could open the door to claims that do.

Why it matters for SEOs and link builders

For SEOs and publishers, the case is about the value of original content in an AI-mediated web. If chatbot answers that reproduce protected work can infringe, then the content that AI systems quote has a legal status as well as a commercial one — and publishers have a lever they did not have before. That strengthens the case for investing in original reporting, data and analysis, because that is the content whose reproduction has value and whose misuse has a remedy.

For link builders, the longer-term implication is about how AI answers attribute their sources. A legal regime that treats reuse as infringement will push AI companies toward clearer attribution, licensing, or both — and attribution is what turns a citation into a link. The November 3 opinion is one step in a long process, but it points in a direction that favours publishers who create material worth citing, and worth linking to.

Sources

Sources: MLex, November 3, 2026; The Gauge, November 3, 2026.

Related reading on this site: Brave Sells Its Search Index to AI — and the New York Times Is Suing, Penske and Chegg Appeal Dismissal of AI Overviews Antitrust Claims, Google, Amazon, Microsoft and Meta Could Dominate AI Agents, Open Markets Warns, Anthropic Reportedly Moving Ahead With IPO Plans, Reddit Takes 29.9% of Gemini Citations in Ahrefs' November Ranking, and Google AI Overviews Cost French Sites Up to 69% of Clicks — But Not Every Site.

More from Linkbaits

Turn this playbook into links

Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.

Start free trial