An adviser to the European Union’s highest court says Google’s Gemini chatbot may infringe publishers’ copyrights when it uses their articles to generate answers. In a non-binding opinion, Advocate General Maciej Szpunar distinguished between AI training and chatbot responses: training models on copyrighted material is likely permitted under the EU’s text and data mining exception, but reproducing expressive content in answers raises a separate issue.
The opinion stems from a case brought by a Hungarian publisher and is the first of its kind to reach the Court of Justice of the European Union (CJEU). Szpunar said copyright protects the specific expression of information, not facts themselves.
The court is expected to issue its final ruling in the coming months. If it follows the advice, Google and other AI developers could have to pay licensing fees for content used in chatbot summaries, potentially affecting traffic to news websites.