🔍 Read the full analysis: What The EU AI Chatbot Copyright Opinion Could Mean For Your Business on ThorstenMeyerAI.com
Get hardware and tech essentials 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 a chatbot can infringe press publishers’ rights if it accesses articles unlawfully or reproduces recognisable expression. He said the available facts in the dispute over Google Gemini do not appear to meet that test. The opinion is nonbinding; the court has not set a date for its judgment.
Advocate General Maciej Szpunar advised the European Union’s top court on Nov. 3 that AI chatbots may infringe press publishers’ rights when they retrieve and reuse articles in certain circumstances. In the dispute over Google’s Gemini chatbot, however, he said the available facts do not appear to satisfy his proposed test—a distinction that could shape how businesses build AI search and answer tools, though the opinion is not a binding ruling.
The case, Like Company v Google Ireland (C‑250/25), began after Hungarian publisher Like Company challenged Gemini’s summary of an article about a proposal to bring dolphins to Lake Balaton. The article was published in July 2023. Like Company argued that Google used its articles to train Gemini and repeatedly reproduced them through chatbot answers between June 2023 and May 2024, according to the Court of Justice’s press release.
Szpunar’s proposed test focuses on two issues: how the chatbot obtained the material and what it returned to the user. An infringement may arise if a system gets around an access restriction, such as a paywall or anti-scraping measure, or if its answer reproduces the article’s distinctive expression so closely that readers can avoid visiting the original. Repeating facts from a report, by itself, would not meet that standard.
On the example before him, Szpunar said the conditions do not appear to be present: he found no indication of access barriers or recognisable wording in Gemini’s answer, and noted that Google appeared to have safeguards. The Hungarian court, which referred the questions to Luxembourg, will decide the dispute after the EU court issues its judgment.
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
The Compliance Line for AI Answers
For businesses offering AI chatbots, search products or web-based assistants, the opinion points to practical areas of copyright risk: whether a system respects restrictions on access, and whether generated answers reproduce protected expression rather than summarise information in new language. Those questions may affect product design, content retrieval, answer length, attribution and licensing discussions. Szpunar’s opinion does not set a final legal rule, but it gives companies a view of the issues the court is being asked to resolve.
For publishers, the proposed approach offers a possible route to challenge systems that obtain articles unlawfully or deliver answers that substitute for the original publication. It does not give publishers control over the underlying facts or a monopoly on news reporting. Businesses should not treat the opinion as a blanket requirement to license every fact used in an AI response, nor as assurance that all summaries are permissible.
Users could see more answers that link to source publications or combine reporting without reproducing distinctive wording, if services respond to legal and commercial pressure in that direction. The practical effect will depend on the court’s eventual judgment, the facts of future disputes and how national courts apply the rules.
Products Worth Considering
Logitech MX Keys S Wireless Keyboard, Graphite
Apple MacBook Pro 16.2-inch with M5 Pro, 18-Core CPU, 20-Core GPU, 48GB Memory, 1TB SSD, Wi-Fi 7, Silver
Ledger Flex Crypto Wallet
As an affiliate, we earn on qualifying purchases.
A 2019 Press Publishers’ Right
The dispute concerns a right established by the EU’s 2019 Copyright in the Digital Single Market Directive. The right covers certain online uses of press publications by online services. As the opinion recounts, lawmakers introduced it with news aggregators such as Google News in mind; the case asks whether and how that framework applies to chatbot answers.
The Budapest-area court sent four questions to the Court of Justice: two about alleged use of articles to train AI models and two about chatbot responses. Nine EU governments and the European Commission filed written observations, and the hearing took place on March 10, 2026, according to the opinion.
Szpunar advised the court to treat the training questions as inadmissible because the record did not establish that Like Company’s articles were used to train Gemini. He reasoned that the specific article appeared after the model’s training data would likely have been collected and that Gemini could have obtained it through live web retrieval. He also said the press publishers’ right concerns online use and, in his view, does not cover model training on dedicated computers.
As an additional observation—not the basis for his procedural recommendation—Szpunar said EU text-and-data-mining rules generally cover AI training in principle, subject to rights holders’ ability to opt out in a machine-readable form. That discussion should not be confused with a finding that Google trained Gemini on Like Company’s material.
“The first generative AI case to reach the Court of Justice.”
— Court of Justice of the European Union, press release No. 155/26
The Court Has Not Ruled Yet
Szpunar is an independent legal adviser, and his opinion is nonbinding. The Court of Justice may adopt his reasoning, modify it or reach a different conclusion. No date for its judgment has been announced, and the Hungarian court will still need to resolve the underlying case.
The available material also does not establish that Like Company’s articles appeared in Gemini’s training data. Szpunar’s comments on text and data mining were made for completeness after he recommended that the training questions be dismissed as inadmissible; they are not a finding about Google’s actual training practices in this case. The scope of any eventual ruling and its application to other AI systems will depend on the court’s final reasoning and the facts before national courts.
Waiting for the Luxembourg Judgment
The next major step is the Court of Justice’s judgment in Case C‑250/25, followed by further proceedings before the Hungarian court. The Court of Justice had not set a judgment date when the opinion was issued. Its statistics show that preliminary references took an average of 16.9 months in 2025, but that figure is a historical average, not a timetable for this case.
Until the judgment arrives, businesses can treat the opinion as a legal signal, not a final compliance rule. They can review whether their systems respect website access controls and whether answers risk reproducing distinctive passages, while tracking the court’s ruling and any later national decisions. The case does not, on its own, determine how every form of AI training or every chatbot response should be treated.
Key Questions
Did the EU court rule that Gemini infringed copyright?
No. The Nov. 3 document is an Advocate General’s nonbinding opinion, not the Court of Justice’s judgment. Szpunar said the available facts in the Gemini dispute do not appear to meet his proposed test.
When could a chatbot answer infringe publishers’ rights?
Under Szpunar’s proposed approach, risk may arise if a chatbot obtains an article by bypassing an access restriction or reproduces its distinctive expression so recognisably that users can avoid the original publication. Reporting the same facts alone would not be enough.
Did the opinion find that Google trained Gemini on the publisher’s articles?
No. Szpunar said the record did not support that claim and recommended that the court dismiss the training questions as inadmissible. His separate discussion of text-and-data-mining rules was general and did not establish that the articles were used in training.
What should businesses operating AI search tools take from the opinion?
The opinion highlights access restrictions and close reproduction of protected expression as issues to monitor. It is not a final ruling or a blanket licensing requirement; companies should follow the Court of Justice’s judgment and assess their systems against the rules that ultimately apply.
Source: ThorstenMeyerAI.com
