An adviser to the EU’s highest court said November 3 that AI chatbots’ retrieval and reuse of news articles can, in some circumstances, infringe publishers’ rights. The opinion is not binding, and no judgment has been issued.
Advocate General Maciej Szpunar’s opinion concerns Like Company v. Google Ireland, C-250/25. The Hungarian publisher challenged Gemini’s use of its journalism; the national court referred questions of EU law.
According to the court’s official summary, Szpunar distinguished protected expression from unprotected news facts. His analysis considers unlawful access, including bypassed technical protections, and answers that recognizably reuse an article’s expression so extensively that readers need not visit the original.
Training is a separate issue. He recommended rejecting those questions as hypothetical here: there was no evidence the articles trained Gemini, and he considered training outside the press-publisher right’s online-use scope. He nevertheless offered the view that training generally falls within the text-and-data-mining exception. Neither position is a court ruling.
The court’s case explainer describes publishers’ concern that AI summaries can divert audiences and revenue. For broadcast newsrooms publishing online, the issue is control over reuse of reporting—not ownership of underlying facts. The opinion creates no immediate payment obligation or U.S. precedent.
Related: Turkey’s separate publisher protections in its Google AI search investigation.