EU Adviser Flags Copyright Risks in Gemini Article Reuse

Maciej Szpunar, an advocate general at the Court of Justice of the European Union (ECJ), issued a legal opinion on the 4th, local time, saying Google’s Gemini chatbot could infringe copyright under certain conditions when it searches news articles and reuses them in answers. Reporting only facts contained in an article is permissible, he said, but infringement may occur if an answer reproduces the original wording or provides enough content that readers no longer need to read the source.

The opinion concerns a lawsuit filed against Google by Hungarian digital publisher Like Company. The publisher alleged Google used its copyrighted works to train Gemini between June 2023 and May 2024 and repeatedly reproduced them in chatbot answers. A Hungarian court asked the ECJ to interpret EU copyright law. Szpunar added that the case had not established evidence that Google used the articles to train its model.

Szpunar distinguished using articles to train a large language model (LLM) from reusing them in chatbot answers. EU text and data mining rules may allow copyrighted material to be used for LLM training, but that does not automatically permit using the material in actual responses, he said. Bypassing paid subscription barriers or technical restrictions to obtain articles, or reproducing recognizable wording from the source in an answer, may also infringe copyright.

EU copyright law protects the specific expression of an article, rather than the simple facts it reports, Szpunar said. His opinion is not legally binding, and the Hungarian court will make the final ruling. The ECJ often follows its advisers’ recommendations, however, so the opinion may affect future disputes between EU publishers and AI search and summary services.

Google welcomed the view that using publicly available information for AI training is permitted under EU law. A company spokesperson said Gemini learns language patterns instead of copying content to create original results, and gives rights holders the option to opt out of AI features. Angela Mills Wade, executive director of the European Publishers Council (EPC), said Google must clearly distinguish search services from AI answers now that it provides users with AI-generated responses. Publishers should not be forced to choose between search visibility and copyright protection, she added.

Maciej Szpunar, an advocate general at the Court of Justice of the European Union (ECJ), issued a legal opinion on the 4th, local time, saying Google’s Gemini chatbot could infringe copyright under certain conditions when it searches news articles and reuses them in answers. Reporting only facts contained in an article is permissible, he said, but infringement may occur if an answer reproduces the original wording or provides enough content that readers no longer need to read the source.

The opinion concerns a lawsuit filed against Google by Hungarian digital publisher Like Company. The publisher alleged Google used its copyrighted works to train Gemini between June 2023 and May 2024 and repeatedly reproduced them in chatbot answers. A Hungarian court asked the ECJ to interpret EU copyright law. Szpunar added that the case had not established evidence that Google used the articles to train its model.

Szpunar distinguished using articles to train a large language model (LLM) from reusing them in chatbot answers. EU text and data mining rules may allow copyrighted material to be used for LLM training, but that does not automatically permit using the material in actual responses, he said. Bypassing paid subscription barriers or technical restrictions to obtain articles, or reproducing recognizable wording from the source in an answer, may also infringe copyright.

EU copyright law protects the specific expression of an article, rather than the simple facts it reports, Szpunar said. His opinion is not legally binding, and the Hungarian court will make the final ruling. The ECJ often follows its advisers’ recommendations, however, so the opinion may affect future disputes between EU publishers and AI search and summary services.

Google welcomed the view that using publicly available information for AI training is permitted under EU law. A company spokesperson said Gemini learns language patterns instead of copying content to create original results, and gives rights holders the option to opt out of AI features. Angela Mills Wade, executive director of the European Publishers Council (EPC), said Google must clearly distinguish search services from AI answers now that it provides users with AI-generated responses. Publishers should not be forced to choose between search visibility and copyright protection, she added.