EU court adviser outlines when AI chatbots may infringe press publishers’ rights
Technology & SocietyWell-coveredThe Facts
- Advocate General Maciej Szpunar said chatbots may infringe press publishers’ rights in certain circumstances.
- The opinion concerns Google’s Gemini chatbot.
- Like Company, a Hungarian website operator, brought the underlying case against Google.
- Szpunar said copyright protects a publication’s expression, not the facts it reports.
- Szpunar said using protected material to train large language models is permitted under the text-and-data-mining exception.
- The advocate general’s opinion does not bind the Court of Justice of the European Union.
- The case is the first concerning generative AI to reach the Court of Justice of the European Union.
Context
What use of news articles could infringe publishers’ rights?
Szpunar said infringement may arise when a chatbot retrieves a press publication from a publisher’s website, reuses it to generate an answer, and supplies that answer to a user. Reuters Economic Times
Does the opinion mean chatbots cannot report news facts?
No. Szpunar said publishers’ rights protect the specific expression of a publication, rather than the facts reported in it. Reuters El Comercio Perú HERALDO
What happens next?
The Court of Justice of the European Union will decide the case. The advocate general’s opinion is advisory and is not binding on the court. La Voz de Galicia Boursorama Straits Times
Where Left and Right agree
Left and right largely agree on this one.
- What both sides accept
- Szpunar's line is correct: training on protected material is permitted under the text-and-data-mining exception, but chatbot outputs infringe if they reproduce a publisher's protected expression rather than facts.
- Where Left and Right differ in emphasis
- The left and the right agree on Szpunar's copyright line — they part on why it matters.
How left and right read it
Information should circulate freely, but the labor that produces it cannot simply be absorbed by whoever owns the largest model — here, a single Hungarian website operator up against Google. So the line Szpunar draws is the right one: training on protected material is permitted under the text-and-data-mining exception, yet what Gemini returns to users can still take a publisher's protected expression rather than the facts it reports.
Property rights do not weaken because the taker is Google. That is why Szpunar's line holds up: copyright protects a publication's expression, not the facts it reports, and training under the text-and-data-mining exception remains permitted — ownership honestly bounded, not expanded. The opinion does not bind the Court. In the first generative-AI case to reach it, the Court should enforce that boundary for Like Company.
Training on a publisher's work can be lawful even when what a chatbot repeats back may still infringe it — the boundary sits at the output, not the input.
The receipts — 75 sources
Wire services (3)
Independent coverage (72)
The news, read by both sides.
The day’s biggest stories in one short brief, built to show both sides and the competing values behind the headlines. Free in your inbox.