A legal dispute involving Google’s Gemini artificial intelligence chatbot has brought a question central to the future of digital publishing before the European Union’s highest court: when does an AI-generated answer become an unlawful substitute for the journalism on which it draws?
The issue has substantial commercial implications for news organisations, technology companies, digital publishers and businesses whose communications depend on online visibility.
On 3 November, Advocate General Maciej Szpunar delivered his opinion in Case C-250/25, Like Company v Google Ireland.
He concluded that AI chatbots retrieving and reusing news publications may, under certain circumstances, infringe rights granted to press publishers under EU law.
The opinion is not binding and the Court of Justice has not yet delivered its judgment.
Nevertheless, it provides an indication of how European copyright rules could be interpreted as generative AI becomes more closely integrated into search and information services.
From Search Results to Generated Answers
Traditional internet search generally presents users with links directing them towards websites containing relevant information.
Publishers have developed business models around this relationship, generating revenue through advertising, subscriptions and commercial partnerships.
Generative AI introduces a different method of accessing information.
An AI chatbot may retrieve information from several sources, combine it and present a direct answer without requiring the user to visit the original websites.
This process can provide users with rapid access to information, but it also creates a commercial question for publishers.
If users obtain sufficiently detailed answers without visiting the original publication, the publisher may lose traffic and associated revenue.
The difficulty lies in determining when the use of published material crosses the boundary between reporting facts and reproducing protected expression.
What the Advocate General Said
The dispute originated with Like Company, a Hungarian operator of online news portals, which brought proceedings against Google over the use of its material by Gemini.
The Hungarian court referred questions concerning both the training of large language models and the generation of chatbot responses.
In his opinion, Szpunar distinguished between these two activities.
He suggested that using protected material for AI training may generally fall within the text-and-data-mining exception provided by EU law.
However, he considered the specific training questions inadmissible in this case because the evidence did not establish that Like Company’s articles had been used to train Gemini.
For chatbot responses, the legal analysis was different.
The Advocate General identified circumstances in which retrieving and reproducing press material could infringe publishers’ rights.
These include unlawful access to protected publications, such as circumventing effective technical restrictions, and recognisable reproduction of protected expression to an extent that substitutes for the original article.
Importantly, copyright does not protect facts themselves.
A chatbot reporting information contained in a news story does not necessarily infringe copyright merely because the same facts appeared in a publisher’s article.
The distinction concerns the protected form of expression rather than exclusive ownership of the underlying information.
The Case Is Not a Victory for Either Side
The opinion contains qualifications that are important for publishers and technology companies alike.
As Reuters reported, the Advocate General’s position does not establish that every AI-generated summary requires authorisation or payment.
Indeed, his preliminary assessment indicated that the specific example submitted by Like Company might not establish infringement, subject to the findings of the national court.
Google has maintained that Gemini generates original responses and has pointed to controls available to publishers.
The case therefore does not establish a general obligation for AI providers to license all material used when generating factual answers.
Nor does it settle the wider debate over the legality of AI training.
The eventual judgment will need to be interpreted within the specific rights, exceptions and circumstances examined by the Court.
What This Means for Publishers
For media organisations, the case reinforces the importance of understanding how their content is accessed and reproduced by AI services.
Publishers may need to review website access controls, content licensing arrangements and the extent to which third-party services reproduce distinctive elements of their reporting.
They must also consider the commercial consequences of restricting automated access.
Blocking certain systems may protect content from unwanted reuse, but it can also affect discoverability and referral traffic.
That creates a difficult balance between protecting intellectual property and remaining visible in digital information services.
Publishers are increasingly dealing with different kinds of automated systems, including conventional search crawlers, AI search products and tools that retrieve content in response to individual user requests.
A single technical policy may not address all these uses.
A Wider Issue for Corporate Communications
The implications extend beyond newspapers.
Companies, professional-services firms and public-affairs organisations increasingly rely on published content to establish expertise and reach prospective clients.
As AI-generated answers become another means of discovering information, businesses have an interest in understanding whether their material is being accurately attributed, summarised or reused.
Communications strategies may need to incorporate monitoring of AI-generated references alongside conventional search analytics.
For organisations commissioning specialist research, reports or original commentary, licensing arrangements and contractual ownership of content may also become more important.
These considerations are particularly relevant where proprietary analysis forms part of a company’s commercial offering.
An Unresolved Commercial Relationship
The forthcoming Court of Justice judgment could influence negotiations between publishers and AI providers.
However, copyright litigation alone cannot resolve every question raised by changes in online information consumption.
Attribution, licensing, audience measurement and the distribution of commercial value are separate issues that may require contractual or regulatory solutions.
The immediate lesson from the Like Company dispute is that the distinction between finding information and reproducing a publication is becoming legally and commercially important.
For publishers and businesses alike, the next phase of digital communications will involve not only ensuring that content can be discovered, but determining the conditions under which other services may reuse it.
