SHOMEDIAOn November 3, Advocate General Maciej Szpunar delivered an opinion in Case C‑250/25, Like Company v Google Ireland, addressing chatbots that retrieve news articles while generating answers. He advised that this practice may, in some circumstances, infringe press publishers' rights. The opinion is not a ruling: the Court of Justice of the EU has yet to decide the case.
The Court's official announcement describes a complaint from Hungarian publisher Like Company, which alleges that Google Gemini extracted and reproduced its journalism without permission. The legal issue is not simply whether a chatbot links to an article, but whether it reuses protectable expression that a reader can recognize in the generated output.
The published opinion distinguishes training a language model from retrieving live articles through retrieval-augmented generation, or RAG. Szpunar recommended treating the specific training questions as hypothetical because the record did not establish that the publisher's articles were used to train Gemini. Reporting that the EU has banned AI training on news would be inaccurate.
For answers generated with retrieved material, the analysis considers whether effective technological access restrictions were bypassed, whether recognizable protected text appears in the answer, and whether the AI provider took reasonable preventative steps. Summarizing a topic or providing a link does not automatically prove infringement.
Publishers and SEO teams can document original publication dates, access controls, licensing terms and concrete examples of recognizable reuse. A robots.txt entry alone is not a complete legal safeguard, and lost referral traffic does not by itself establish a copyright violation.
Any wider effect on content licensing and AI-powered discovery will depend on the Court's final judgment. This preliminary opinion is an important development for news media, but not a new binding rule that has already resolved publishers' claims.
Комментарии