🔍 Read the full analysis: Publishers’ Rights And AI Chatbots: What The EU Opinion Says on ThorstenMeyerAI.com
Get smart everyday buys 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 Nov. 3, 2026, Advocate General Maciej Szpunar advised the EU Court of Justice that chatbots retrieving and reusing news articles may infringe press publishers’ rights in certain circumstances. His proposed test focuses on unlawful access and answers that reproduce recognisable expression; he said the test does not appear to be met in the dispute over Gemini’s summary of a Hungarian article. The opinion is not binding, and the court has not set a judgment date.
Advocate General Maciej Szpunar advised the European Union’s Court of Justice on Nov. 3, 2026, that AI chatbots which retrieve and reuse online news articles may, in certain circumstances, infringe press publishers’ rights. The opinion in Case C-250/25 sets out a proposed test focused on whether a chatbot bypassed access restrictions or reproduced an article’s recognisable expression, but Szpunar said those conditions do not appear to apply to the Gemini summary at the heart of the case.
The dispute began after Gemini summarised an article from a Hungarian news portal about a singer who reportedly wanted to bring dolphins to Lake Balaton. The portal’s publisher, Like Company, sued Google in Hungary, alleging that Google used its articles to train Gemini and reproduced and made them available through chatbot answers between June 13, 2023, and May 31, 2024. Those are the publisher’s allegations, not findings established by the opinion.
Szpunar’s central view is that a chatbot answer can infringe publishers’ rights if the service obtained an article unlawfully—for example, by circumventing a paywall or anti-scraping measure—or if it repeats the article’s form of expression so recognisably that a user can avoid visiting the original publication. Repeating the same facts, by itself, would not meet that test. In this case, he said, the example answer did not appear to contain recognisable wording, the article was not behind an access barrier, and Google appeared to have safeguards.
The opinion treats the training allegations separately. Szpunar advised the court to find the two questions about AI training inadmissible because the record did not establish that Like Company’s articles were used to train Gemini. He also considered training “for the sake of completeness,” saying text and data mining rules generally cover it in principle, subject to the possibility of rights holders opting out in a machine-readable form. The court, not the Advocate General, will decide the case.
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 proposed test could help define how a copyright rule created for online uses of press publications applies to newer services that generate answers from web content. It distinguishes between using reported facts and copying a publisher’s protected expression, while making the way a chatbot accesses an article relevant too. That distinction matters to publishers, AI providers and users seeking news through conversational search.
If the court adopts the approach, publishers would not gain control over the facts in their reporting simply because an AI system repeats them. They could, however, have grounds to challenge answers that draw on material accessed by bypassing restrictions or that reproduce distinctive expression closely enough to substitute for the original article. For chatbot operators, the opinion points to the importance of respecting access controls and avoiding answers that function as near-copies.
The opinion may also shape commercial discussions over licensing and safeguards, but it does not itself settle anyone’s legal liability. Any practical effect depends on the final judgment and how national courts apply the rules to evidence about particular systems and answers.
Products Worth Considering
Apple MacBook Pro 2021 14-inch with M1 Max Chip, 64 GB RAM, 2 TB SSD, Silver
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
HP OfficeJet 250 Wireless Mobile All-in-One Inkjet Printer
As an affiliate, we earn on qualifying purchases.
How the Hungarian Case Reached Luxembourg
Like Company operates several Hungarian web portals and brought its claim before the Budapest regional court. The claim relies on Article 15(1) of the EU Copyright in the Digital Single Market Directive, adopted in 2019. That provision grants press publishers rights over certain online uses of their publications. The Advocate General’s opinion recalls that lawmakers had news aggregators such as Google News in mind when creating the right.
The Hungarian court referred four questions to the Court of Justice: two concerning AI training and two concerning chatbot answers. According to the opinion, nine EU governments and the European Commission submitted written observations, and the court held a hearing on March 10, 2026. The reference asks whether a legal framework developed before today’s widespread generative chatbots reaches the way those systems use news content.
On training, Szpunar’s main conclusion was procedural, not a final ruling that every form of AI training is lawful. He said the record did not support the claim that the article was part of Gemini’s training data; the article appeared in July 2023, while Gemini 1.0 was released in December 2023. He considered live retrieval from the website a more plausible explanation for the summary. He also reasoned that training on dedicated computers is not the “online use” covered by the press publishers’ right. Separately, his discussion of the text and data mining exception was supplementary to his advice that the training questions were inadmissible.
“Google “repeatedly reproduced and made [its articles] available to the public through the Gemini chatbot.””
— Like Company, as described in the Court of Justice press release
Key Questions Await the Court
The Advocate General’s opinion is not binding. The Court of Justice may follow it, modify its reasoning or reach a different conclusion. No date for the judgment has been announced, so the final legal test and its application to Google remain unsettled.
The training issue is also not resolved on the merits by this opinion. Szpunar recommended that the court reject those questions as inadmissible because the case materials did not show the articles were used for training. His additional discussion of text and data mining does not establish what happened with Gemini’s training data or decide whether a particular rights holder validly opted out.
The Hungarian court will ultimately need to assess the evidence in the dispute. The opinion’s account that Google appeared to have safeguards and that the example answer lacked recognisable wording is a provisional assessment, not a final finding of fact or a ruling that chatbot summaries generally comply with EU law.
Court Judgment and Case Return
The Court of Justice will issue its judgment in the preliminary-reference case; no timetable has been set. Once it rules, the Hungarian court will apply the EU court’s interpretation to Like Company’s claims and the facts before it. The eventual judgment will determine whether the publisher’s case can proceed under the press publishers’ right and how the proposed distinction applies here.
Until then, the opinion offers guidance rather than a binding decision for publishers and AI services. The decisive questions remain whether access restrictions were bypassed, whether chatbot outputs reproduce protected expression, and what evidence can establish how a system obtained and used particular articles.
Key Questions
Has the EU Court of Justice ruled that Gemini infringed publishers’ rights?
No. Szpunar issued a non-binding opinion, not the court’s judgment. He said the proposed conditions for infringement did not appear to be met in the example discussed, but the Court of Justice has yet to rule.
When could a chatbot answer infringe under the opinion?
Szpunar’s proposed test covers cases where a chatbot obtains an article by bypassing access restrictions, or where its answer reproduces the article’s recognisable expression so that users can avoid accessing the original. Repeating the same facts alone would not be enough under that test.
Did the opinion decide whether AI training on news is lawful?
No. Szpunar advised that the training questions should be found inadmissible because the case materials did not show that the articles were used to train Gemini. His additional comments on text and data mining were not a final decision on the legality of this system’s training.
What happens next in the case?
The Court of Justice will issue a judgment, with no date announced. The Hungarian court will then apply the EU court’s interpretation to the dispute between Like Company and Google.
Source: ThorstenMeyerAI.com
