🔍 Read the full analysis: The EU’s Latest AI Copyright Question: Can Chatbots Infringe Rights? on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
On November 3, 2026, Advocate General Maciej Szpunar advised the EU Court of Justice that chatbots can infringe press publishers’ rights if they obtain articles unlawfully or reproduce recognisable expression. He said the test does not appear to be met in Like Company’s dispute with Google over a Gemini summary, and the opinion is not binding.
The EU Court of Justice’s Advocate General, Maciej Szpunar, said on November 3 that chatbots may infringe press publishers’ rights when they retrieve and reuse news articles in answers, but only under specific conditions. In a dispute involving Google’s Gemini and Hungarian publisher Like Company, he said the available facts do not appear to satisfy those conditions; the court has not yet ruled.
In his opinion in Case C‑250/25, Szpunar set out two circumstances in which a chatbot’s use of an article could infringe the EU press publishers’ right. One is where the service obtained the article unlawfully, such as by bypassing a paywall or an anti-scraping measure. The other is where the chatbot reproduces the article’s expression so recognisably that a user can dispense with visiting the original publication.
By contrast, repeating facts reported in an article is not enough on its own, the Advocate General said. The distinction is between information, which is not protected by copyright as such, and the article’s particular wording or other protected expression. Applying that test to the example in the case, Szpunar said there appeared to be no access barrier, no recognisable wording in Gemini’s summary and safeguards at Google. He therefore said the conditions for infringement did not appear to be met.
Like Company, which operates Hungarian news websites, brought the case in a Budapest court. It alleges that Google used its articles to train Gemini and repeatedly reproduced or made them available through the chatbot from June 13, 2023, to May 31, 2024. The Budapest court referred four questions to Luxembourg. Szpunar advised that the two questions about training were inadmissible because the record did not show that the publisher’s articles had been used to train the model. His opinion is advisory, not binding; the Court of Justice will issue the final ruling.
Infographic 01 · Where the line runs
When a chatbot answer infringes a publisher's right
Advocate General Szpunar's test in Case C‑250/25 for chatbots that retrieve news articles in real time and reuse them in their answers.
Stays free
- Reporting the same facts. Nobody owns the news itself (pts 199–200).
- A short factual summary, presented differently (pt 208).
- An overview built from several sources, restructured, with links back (pts 208, 234).
- Individual words or very short extracts (pt 201).
Can infringe
- Getting around a paywall or anti-scraping block to fetch the article (pt 188).
- An answer that keeps the article’s structure and phrasing, even paraphrased, so readers can skip the original (pts 202, 205, 207).
- Systematically recognisable articles in answers to related questions: making available to the public (pts 213, 240).
The test, step by step
For a chatbot that fetches a news article live (RAG) and answers a user
- Is it an online use by a service in the EU?Yes for RAG answers to EU users, wherever the servers are (pts 165–169).
- Did the bot get in lawfully?No circumvention of effective technical measures (pt 188).
- Is the article’s form of expression recognisable in the answer, to the point of substitution?Side-by-side comparison; facts do not count (pts 196–209).
- Is the provider responsible?If inherent to its business model, or no reasonable measures such as respecting blocks and output filters (pts 221–222).
- Temporary copies during retrieval are then exempt under Art. 5(1) InfoSoc; the TDM exception does not apply to RAG (pts 179–189).
An Advocate General’s opinion, not a judgment. General information, not legal advice.
Sources: Opinion of AG Szpunar, C‑250/25 (provisional text), CURIA press release No 155/26 (3 Nov 2026). Point numbers refer to the provisional text.
Powered by Thorsten Meyer AI
A Narrower Test for Chatbot Answers
The opinion offers a legal framework for publishers and AI services as chatbots increasingly answer questions using material found online. Under Szpunar’s proposed test, publishers would not gain control over the underlying facts in their reporting, but they could have a claim if a service crosses access restrictions or its answer functions as a substitute for the article by retaining recognisable expression.
For AI companies, that points to the importance of respecting paywalls and anti-scraping controls and avoiding outputs that closely reproduce protected wording. For publishers, the opinion could support licensing arguments where a chatbot uses material in those ways, while stopping short of establishing a general right to charge for every answer based on reporting. These are potential implications of the advice, not a final legal outcome.
The distinction also matters to users. If the court adopts the approach, answers that synthesise facts from articles may remain permissible, while near-copies or material obtained by evading access limits could face legal challenge. The precise effect will depend on the court’s judgment and how national courts apply it to particular systems and outputs.
Products Worth Considering
LG UltraGear 32GX850A-B 32-inch 4K UHD Glossy OLED Gaming Monitor
Samsung Odyssey G4 25-inch FHD Gaming Monitor
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
As an affiliate, we earn on qualifying purchases.
A Dispute Over Gemini’s Summary
The case began with a July 2023 Hungarian article about a singer’s reported plan to bring dolphins to Lake Balaton. A user later asked Gemini to summarise the article in Hungarian, and the publisher took the response to court. The dispute concerns the EU press publishers’ right established by the 2019 Copyright Directive, which allows publishers to control certain online uses of their press publications by services.
That right was designed with online services such as news aggregators in mind. The case asks whether it also applies when a chatbot retrieves articles in response to a user’s question and produces an answer. The referring court sent four questions to the EU Court of Justice: two concerning AI training and two concerning chatbot responses. The hearing took place on March 10, 2026, after written observations from nine EU member states and the European Commission.
On training, Szpunar’s main conclusion was procedural: the questions should not be answered because the case materials did not establish that Like Company’s articles were included in Gemini’s training data. He considered it more plausible that Gemini had accessed the article live through retrieval-augmented generation, a process in which a system retrieves online material and uses it to formulate an answer. Separately, he offered a preliminary view that EU text-and-data-mining rules may generally cover AI training, subject to rights holders’ ability to opt out in a machine-readable form. That discussion was not his ruling on the dispute’s training questions.
The Court Has Yet to Decide
The Advocate General’s opinion is not the final judgment. The Court of Justice may follow his reasoning, modify it or reach a different conclusion. No date for the judgment has been announced, and the Hungarian court will have to decide the dispute after the EU court answers the referred questions.
The case record, as described in the opinion, did not establish that Like Company’s articles were used in Gemini’s training data. That does not resolve broader disputes about training other models on copyrighted material. Nor does the opinion settle how courts should assess varied chatbot outputs, access methods, or the level of similarity that makes an answer a substitute for an original article.
Szpunar’s comments on the text-and-data-mining exception are also not a binding ruling on training. They were offered after he advised that the questions were inadmissible, and the court has not yet adopted that view. The scope of any rights-holder opt-out, including how it must be expressed and applied in specific cases, remains subject to legal interpretation.
A Binding Judgment Still Pending
The Court of Justice will now consider the case and issue its judgment. Its ruling will answer the questions referred by the Budapest court and guide the national proceedings. The court has not announced a timetable; its 2025 statistics put the average duration of preliminary-reference proceedings at 16.9 months, though that figure does not predict when this case will be decided.
After the EU court rules, the Hungarian court will apply the interpretation to the evidence in Like Company’s dispute with Google. Until then, Szpunar’s proposed test is guidance rather than binding law. Publishers, AI providers and users will be watching whether the final judgment keeps the distinction between reporting facts and reproducing an article’s protected expression, and how it treats access restrictions and chatbot safeguards.
Key Questions
Has the EU Court of Justice ruled that Gemini infringed copyright?
No. Advocate General Maciej Szpunar issued a nonbinding opinion. He said the conditions for infringement did not appear to be met on the facts described, and the court’s judgment is still pending.
When could a chatbot answer infringe a publisher’s rights?
Under Szpunar’s proposed test, infringement may arise if a chatbot obtains an article unlawfully, such as by bypassing an access restriction, or reproduces its expression so recognisably that users can do without the original publication. The Court of Justice has not yet adopted this test.
Did the opinion decide whether AI training on news articles is lawful?
No. Szpunar advised that the training questions in this case were inadmissible because the record did not show the publisher’s articles were used to train Gemini. His additional discussion of the text-and-data-mining exception was preliminary and is not a binding decision on AI training generally.
What happens after the Advocate General’s opinion?
The Court of Justice will issue a judgment interpreting the EU law questions, after which the Hungarian court will decide the case. No judgment date has been announced.
Source: ThorstenMeyerAI.com
