Google’s artificial intelligence chatbot Gemini could breach European Union copyright protections when it uses news articles to generate responses, according to a legal opinion issued Tuesday that could have significant implications for the relationship between AI companies and digital publishers.
Maciej Szpunar, an advocate general at the Court of Justice of the European Union, concluded that AI services may infringe publishers’ rights when they retrieve and reproduce protected journalistic material without authorization, Reuters reported on November 3. The opinion draws a distinction between using copyrighted material to develop AI models and incorporating protected content into answers delivered to users.
The assessment comes as news organizations confront growing competition from generative AI platforms that provide direct answers to information requests, potentially reducing the need for readers to visit publishers’ websites.
A decision that establishes liability for certain uses of news content could strengthen publishers’ positions in negotiations over licensing and compensation, according to Reuters.
Legal Dispute Puts AI-Generated News Under Scrutiny
The opinion stems from a lawsuit brought against Google by Like Company, a Hungarian operator of online news portals.
The publisher alleges that Google used its articles to train the large language models behind Gemini and subsequently reproduced protected material through the chatbot. The disputed activity covers a period between June 2023 and May 2024, according to documents published by the EU court.
A Hungarian court referred questions about the application of European copyright law to the Luxembourg-based tribunal, seeking clarification on the protections available to publishers when AI systems process and reproduce their work.
Szpunar’s assessment indicates that copyright infringement may arise when chatbots obtain articles by circumventing paywalls or other access restrictions. Legal concerns may also emerge when AI-generated responses reproduce recognizable elements of a publication’s protected expression.
The distinction is important because European copyright protections do not grant publishers exclusive ownership over news events or factual information. Instead, they protect the particular expression used to communicate that information, according to Reuters and the court’s published opinion.
AI Training Receives Different Legal Treatment
While the opinion raises potential liability for chatbot responses, it takes a more permissive position on the use of copyrighted material in developing AI models.
Szpunar indicated that training large language models on protected works is generally allowed under the EU’s existing text-and-data-mining exception, Reuters reported.
Related: US Judge Dismisses Publisher Antitrust Cases Against Google Over AI Search
He also recommended that the court decline to address the specific training-related questions in the Hungarian dispute because evidence had not established that Like Company’s articles were used to train Gemini.
That distinction could prove consequential for technology companies developing AI systems that combine previously trained models with real-time information retrieved from the internet.
Gemini uses a process known as retrieval-augmented generation, which allows the chatbot to consult online sources when preparing responses. The technology can improve the relevance of answers, including those involving recent news developments, but also raises questions about how much protected material may be incorporated into the resulting output.
Publishers Face Growing Pressure From AI Summaries
The legal dispute highlights an increasingly important commercial question for the media industry: whether publishers should receive compensation when AI services use their reporting to answer questions without directing readers to the original articles.
Traditional search engines generally present links that users can follow to publishers’ websites. Generative AI systems can instead deliver consolidated answers within their own interfaces.
That change has raised concerns among news organizations about declining referral traffic and the potential consequences for advertising and subscription revenue, Reuters reported.
A court decision supporting publishers’ rights in certain circumstances could provide additional leverage for media companies seeking licensing agreements with AI developers. It could also influence how chatbot operators retrieve, summarize and present material from third-party websites.
Final Court Decision Still Pending
Szpunar’s opinion is advisory and does not constitute a binding judgment. Although judges at the EU’s highest court frequently follow recommendations from advocates general, they are not required to do so.
The court must still determine how European copyright rules apply to the questions referred by the Hungarian judiciary. The underlying dispute would then proceed through the national legal process.
For Google and other developers of generative AI, the outcome could help establish a clearer boundary between permissible use of factual information and unauthorized reproduction of publishers’ protected work.
For news organizations, the stakes extend beyond a single chatbot. The case could influence the economic and legal terms under which journalism is distributed through AI-powered information services across the European Union.
The opinion leaves one central issue unresolved: how AI companies can provide comprehensive answers based on published reporting while respecting the rights of the organizations that produce it.
Source: Reuters
