🔍 Read the full analysis: An EU Court Adviser’s Take On AI Chatbots And Copyright Explained on ThorstenMeyerAI.com
Get audio and creator gear 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 news articles may infringe publishers’ rights in certain circumstances. He said the example involving Google’s Gemini does not appear to meet his proposed test, but the opinion is nonbinding and the Hungarian court will decide the case.
Advocate General Maciej Szpunar advised the EU Court of Justice on Nov. 3 that AI chatbots retrieving and reusing online news articles may infringe press publishers’ rights in certain circumstances. In Case C‑250/25, Like Company v Google Ireland, he said infringement may arise if a chatbot circumvents access restrictions or reproduces a publication’s distinctive expression closely enough that readers can do without the original. His opinion is not binding, and he said the available facts do not appear to satisfy that test in this case.
The dispute began after Hungarian publisher Like Company sued Google in Hungary. The publisher said Google used its articles to train Gemini and repeatedly reproduced and made them available through the chatbot between June 13, 2023, and May 31, 2024, according to the Court of Justice’s press release. The Hungarian court referred four questions to Luxembourg, covering alleged use in training and the chatbot’s answers.
The example at the center of the opinion involved a Hungarian article published in July 2023 about a singer’s reported plan to bring dolphins to Lake Balaton. When a user asked Gemini to summarize the article, the chatbot returned a summary in Hungarian. Szpunar’s proposed test focuses not on whether the answer reports the same facts, but on how the article was accessed and whether its protected wording or other expression is recognisable in the answer.
Szpunar said the training questions should be treated as inadmissible because the record did not establish that Like Company’s articles were used to train Gemini. He reasoned that the dolphin article appeared after the training data for Gemini 1.0 was likely collected, and that the example was more consistent with live web retrieval than training. Separately, he said the press publishers’ right concerns online use of publications and does not cover model training on dedicated computers. His additional discussion of training was not the basis for his recommendation on those questions.
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
Limits on Chatbot Reuse of News
The opinion offers a possible legal boundary for services that search the web and generate answers from news coverage. Under Szpunar’s approach, a chatbot would not infringe the press publishers’ right merely because it conveys the same underlying facts as an article. The focus would be on unauthorised access—such as bypassing a paywall or anti-scraping measure—or on reproducing the article’s form of expression so extensively that users can dispense with the original.
For publishers, that could support complaints or licensing discussions where a service crosses either boundary, while leaving ordinary reporting of facts outside the proposed test. For chatbot providers, the opinion points to the importance of respecting technical access restrictions and avoiding answers that closely reproduce protected material. These are potential implications of an adviser’s opinion, not requirements imposed by a final court judgment.
Readers may also be affected if services respond with more synthesis across sources or links back to original reporting. The opinion does not determine how any specific product must change, however. That will depend on the final ruling and, in this dispute, the Hungarian court’s assessment of the evidence.
Products Worth Considering
DJI Osmo Mobile 7P Phone Gimbal with Mic Mini Transmitter, Infinity Black
Bose SoundLink Plus Portable Bluetooth Speaker, Wireless Outdoor Speaker, Up to 20 Hours Battery Life, Waterproof and Dustproof, Black
DJI Osmo Mobile 8 Gimbal with Mic Mini Transmitter, Infinity Black
As an affiliate, we earn on qualifying purchases.
A Publisher Right Tested by Chatbots
The dispute concerns the EU press publishers’ right under Article 15(1) of the 2019 Copyright in the Digital Single Market Directive. The right was introduced to let publishers control or license certain online uses of press publications. Szpunar’s opinion notes that the legislation was directed in part at news aggregation services such as Google News; the case asks whether and how that right applies when a chatbot answers a user by drawing on online articles.
The Hungarian court sent four questions to the Court of Justice. Nine EU governments and the European Commission submitted written observations, and the hearing took place on March 10, 2026, according to the opinion. The Court of Justice describes the case as its first generative AI case.
On training, Szpunar also discussed the EU’s text and data mining exception, which can permit certain uses of works for analysis, subject to conditions including rights holders’ ability to reserve their rights in a machine-readable form. He said training may fall within that exception in principle, but this was an additional discussion after he recommended that the training questions be dismissed as inadmissible. It is not a final ruling on whether Gemini used the publisher’s works for training.
“The opinion is not binding on the Court of Justice.”
— Court of Justice of the European Union, Press Release No 155/26
The Final Test Is Still Pending
The Court of Justice has not issued its judgment, and no date has been set. The Advocate General’s assessment is independent legal advice, not a decision resolving the dispute. The Hungarian court will ultimately determine the case in light of the EU court’s ruling and the evidence before it.
Several factual and legal questions remain open. The source material does not establish that Like Company’s articles were used to train Gemini; that allegation is part of the publisher’s case, not a confirmed finding. The court adviser also did not decide that every kind of training is permitted: his discussion of the text and data mining exception was qualified as applying in principle and came after his procedural recommendation on the training questions.
It is also not yet clear how the Court of Justice will define recognisable expression in chatbot answers, what level of similarity would allow a user to substitute an answer for the original, or how access restrictions and safeguards will be assessed across different services.
Court Judgment and Hungarian Proceedings
The next major step is the Court of Justice’s judgment in Case C‑250/25. The court has not announced when it will deliver that ruling. Its 2025 statistics put the average time for preliminary-reference cases at 16.9 months; that is a historical average, not a timetable for this case.
After the EU court answers the Hungarian court’s questions, proceedings in Hungary are expected to continue, with the national court applying the EU interpretation to the dispute. Until then, the adviser’s proposed limits can inform debate and legal planning, but they do not settle whether Google infringed Like Company’s rights.
Key Questions
Did the EU Court of Justice rule that Gemini infringed copyright?
No. Szpunar issued a nonbinding opinion, not a judgment, and said the conditions for infringement do not appear to be met on the facts described. The Court of Justice still has to rule.
When might a chatbot’s use of a news article infringe publishers’ rights?
Under the adviser’s proposed test, possible infringement could arise if a chatbot obtains an article by bypassing access restrictions or reproduces its distinctive expression so extensively that users can do without the original. Reporting the same facts alone would not be enough under that test.
Did the opinion find that Gemini trained on Like Company’s articles?
No. Szpunar said the record did not support that claim and recommended treating the training questions as inadmissible. The opinion describes the example as consistent with live retrieval from the web, rather than use of the article in model training.
Does the opinion settle whether AI training is allowed under EU law?
No. The adviser discussed the text and data mining exception as applying to training in principle, subject to conditions, but that discussion was not a binding ruling on the issue. The Court of Justice has not yet decided the case.
Source: ThorstenMeyerAI.com
