EU Advisor Says Google Gemini May Infringe News Copyrights
An EU Court Advocate General stated that Google's Gemini may infringe copyrights when using news reports to generate answers under specific conditions. The opinion distinguishes between protected expression and unprotected facts, while noting that training LLMs on copyrighted material is currently permitted. An adverse ruling could force tech giants to pay for content, reshaping media traffic dynamics.
Full Report
IT Home reported on November 3 that, according to Reuters, an advisor to Europe’s highest court stated on Tuesday that Google’s Gemini chatbot may violate copyright rules in certain circumstances by using news reports when answering questions from internet users.
IT Home noted that this opinion was issued by Maciej Szpunar, Advocate General at the Court of Justice of the European Union (CJEU) based in Luxembourg. It relates to a lawsuit filed by Hungarian website operator Like Company against Google, with the dispute centering on how Google’s AI chatbot, Gemini, uses news content.
If the court ultimately rules against Google and requires the tech giant to pay for the use of media content, it could offer new survival opportunities for online publishers like Like Company. As AI chatbots such as Gemini increasingly replace traditional search engine results—which typically include hyperlinks to original news articles—with generated news summaries, web traffic to online publishers continues to decline.
According to a statement released by the CJEU, Szpunar said: “When a chatbot powered by a large language model (LLM) retrieves a report from a publisher’s website, uses it to generate a response, and provides that response to the user, it may infringe upon the rights enjoyed by the publisher under certain circumstances.”
He pointed out that if the chatbot accesses news content by illegally bypassing paywalls or other technical restrictions, or if it reuses specific expressive forms of news reports in an identifiable manner within its responses, then answers generated by Gemini could constitute infringement of the publisher’s reproduction rights.
However, he also emphasized that copyright rules do not protect “mere facts” contained in news reports, but only the specific expressive forms used to convey those facts. In other words, the facts themselves in news reports are not protected by such copyright laws, but the journalists’ specific textual expression of those facts may be.
On the other hand, Szpunar argued that using copyrighted material to train large language models is “permitted” under current text and data mining exceptions. Large language models are a class of AI systems that learn from massive amounts of text data to generate human-like responses and summarize information.
It should be noted that opinions issued by Advocates General at the CJEU are not legally binding, but judges of the court often refer to and adopt such opinions.