On November 3, 2026, Advocate General Maciej Szpunar delivered his opinion in case C-250/25, Like Company v Google Ireland, which the Court of Justice of the European Union calls the first generative AI case to reach it. In his view, a chatbot that retrieves an online article to answer a user can infringe the press publisher's right in two cases: it circumvented a technical measure restricting access, such as a paywall, or its answer reuses the form of expression of the article to the point that the user no longer needs to read it. Facts stay free. The opinion does not bind the Court, which has not ruled yet. The press publishers' right does not apply to a company website. The opinion does, according to PPC Land, treat accepting crawling for search and consenting to chatbot use as two different things.
Contents
01The case and the Advocate General's two conditions
The right at issue is the press publishers' right, created by Article 15 of the 2019 European directive on copyright in the Digital Single Market.12 Like Company, which operates several Hungarian web portals, invoked it before the Budapest Regional Court, which referred four questions to the Court of Justice.1
The questions put to the Court
Two questions concern training a large language model on press articles: does it fall under the publisher's exclusive right of reproduction, or under the text and data mining exception? The other two concern the chatbot's answers: can they fall under the reproduction right or the right of making available to the public?1
The Advocate General advises the Court to declare the two training questions inadmissible, since there is no evidence that articles were used to train Gemini. He adds, for completeness, that using protected works to train a model is, as a rule, freely permitted under the text and data mining exception.1 He asks the Court to focus on the chatbot's answers, the only ones relevant to the dispute.1
The two conditions
For the answers, the Advocate General retains two factors. The first: did the chatbot access the article unlawfully, in particular by circumventing effective technical measures, such as a paywall, designed to restrict access to the publication or its scraping by bots? The second: does the answer merely report facts contained in the article, or does it reuse its form of expression in a recognizable manner, to the extent that the user, by reading it, is entirely dispensed from accessing the original article?1
PPC Land specifies that the opinion refers to a paywall or a measure designed specifically to prevent access or scraping by bots.2 The right of making available to the public is, as a rule, infringed where the form of expression of the article is systematically recognizable, to the extent described above, in the answers to related queries.1
Facts stay free
The Court's press release sets the limit: the press publishers' right covers the form of expression of publications. The facts they report stay free. Its aim is to protect the investment that goes into making a publication, without giving publishers a monopoly over the news itself.1
According to PPC Land, which cites the opinion, paraphrase is no safe harbor. An answer that reflects the arrangement, structure and general phrasing of an article can reproduce it despite rewording. A short summary of the factual content presented differently, or a general outline built from several sources, is not a reproduction.2 Long analytical articles are by nature more recognizable than purely informative reports.2
Who answers for the use
The chatbot provider is liable for such uses where they are inherent to its business model, or where it has failed to put in place appropriate measures to counter them credibly and effectively.1 According to PPC Land, the opinion places primary liability with the user who directs the chatbot, and expects the provider to program its retrieval bots to respect publishers' access restrictions and to keep its answers away from any specific publication through output filters.2
02Crawling for search and use by a chatbot
Google argued before the Court that a publisher who leaves its articles freely accessible, with no paywall or measure against scraping, implicitly consents to their use in a chatbot's answers. According to PPC Land, the Advocate General writes that he does not think so.2
The reasoning on blocking bots
According to PPC Land, the opinion notes that the bots that ground a chatbot's answers are not fundamentally different from those that index sites for search engines, and that at Google the same bots seem to serve both functions.2 Like Company explained at the hearing that blocking them through robots.txt can affect how the site appears in search. A publisher can therefore accept crawling to stay visible without it being reasonable to conclude that he or she consents to the use of the publications to ground chatbots.2
Google replied that Google-Extended allows this differentiation with no effect on visibility in Google. The opinion finds that this would need to be verified and that, even if true, concluding that there is implicit consent seems far-fetched at the current stage of AI and chatbots.2
On the absence of a paywall, the opinion notes that a publisher can forgo one because it prefers the advertising revenue of broader traffic, which shows no agreement to the reuse of its publications by third parties.2
robots.txt as a reservation of rights
The opinion also addresses the text and data mining exception, which a publisher can set aside through a machine-readable reservation of rights. According to PPC Land, the robots.txt protocol is one, as is Google-Extended, which Google released on September 28, 2023, and metadata embedded in files.2 Per-bot settings, Google-Extended included, are described in the technical configuration of a site for AI crawlers.
PPC Land notes that the record before the Court predates two developments that the opinion does not take up: the Search Console setting Google introduced in June 2026, which excludes a site from AI Overviews and AI Mode without touching conventional ranking, and the shared token announced on October 30 to opt out of training.2 What that token covers and what it leaves out is detailed in the article on the shared robots.txt token.
03What the opinion leaves open for a business
An Advocate General's opinion proposes a legal solution to the Court, in complete independence. The judges are now beginning their deliberations and will give judgment at a later date.1 According to PPC Land, they often follow the Advocate General, though not always.2
The Like Company example
The contested answer summarized an article from the balatonkornyeke.hu site dated July 21, 2023. According to PPC Land, the opinion finds that this example does not appear to meet the test: the site does not seem to have put any technical measure in place against access or scraping, the form of expression of the article is not apparent in Gemini's answer, which essentially gives the same facts in different terms with a single sentence from the article, and Google seems to have put appropriate measures in place.2 The Hungarian court will decide, including whether a portal like Like Company's is a press publication at all.2
The situations at a glance
The table lists the cases that the Court's press release and the opinion, as PPC Land reports it, distinguish.
| Situation | What the opinion says |
|---|---|
| The chatbot circumvents a paywall or a measure against bots | access is unlawful, an infringement of the publisher's right is possible |
| The answer reports facts | facts stay free |
| Short factual summary, presented differently | not a reproduction |
| Answer that reflects the arrangement, structure and phrasing of an article | reproduction possible despite rewording |
| Answer that makes reading the article unnecessary | decisive test of substitution |
| Training a model | as a rule permitted under the text and data mining exception |
What a business that publishes takes from it
The right at issue is the press publishers' right, and it does not apply to a company website. What a small business finds in the opinion is the logic of opting out: robots.txt as a reservation of rights, the distinction between crawling for search and use by a chatbot, the gap between a fact reused freely and a recognizable wording.
The Advocate General also describes the web's historical bargain: search engines copied sites and sent back traffic that publishers could monetize, and that bargain is called into question when chatbots and services such as AI Overviews provide the information directly.2 Measuring that shift on your own site goes through tracking the impact of AI Overviews in Search Console.