Wikimedia Foundation Weighs In on NYT v. OpenAI: Split Training From Chatbot Answers

What happened

The brief comes from the Wikimedia Foundation, which operates Wikipedia and other open knowledge projects, and was filed with the US District Court for the Southern District of New York in In re: OpenAI, Inc. Copyright Infringement Litigation; an amicus brief is filed by a non-party for the court to consider and does not bind the court.

The brief splits the fair use question in two: model training, and retrieval-augmented generation (RAG) and newer techniques serving a similar grounding purpose; for training, it argues that using copyrighted content to train a large language model can be fair use when the resulting model is something new and different from the original sources, with no licensing terms to meet; for RAG, it argues that a chatbot retrieving content from the web and delivering it back to users should not count as fair use and must respect each platform's or publisher's licensing requirements.

Foundation CEO Bernadette Meehan said in the announcement: “We are taking a position in this case to defend the free and open human knowledge that Wikipedia depends on.” The Foundation also states that Wikipedia content is distributed under the Creative Commons CC BY-SA 4.0 license, which requires credit to the original creators and requires adaptations to remain under the same license.

Key facts

1. Filer and date: the Wikimedia Foundation, Oct 30, 2026, filed with the US District Court for the Southern District of New York.
2. Case scope: 12 combined OpenAI copyright cases, titled In re: OpenAI, Inc. Copyright Infringement Litigation, the best known being The New York Times Co. v. Microsoft and OpenAI.
3. The New York Times' claim: filed in 2023, alleging that OpenAI and Microsoft used Times-owned copyrighted content without permission, including for AI model training, in direct violation of copyright law.
4. OpenAI and Microsoft's position: training and deploying models based on publicly available data constitutes technological development and falls within fair use.
5. The Foundation's stance: it takes neither side, saying both parties' readings of fair use are overbroad and simplified.
6. Training rule: using copyrighted content to train a large language model can be fair use when the resulting model is something new and different from the original sources, with no licensing terms to meet.
7. Retrieval rule: RAG and similar grounding techniques that retrieve content from the web and deliver it to users should not count as fair use and must respect each platform's or publisher's licensing requirements; Wikipedia content is licensed CC BY-SA 4.0, requiring attribution and requiring adaptations to remain under the same license.
8. Ruling status: the court has not ruled on the central fair use question in this case, and the brief itself does not bind the court.
9. Wikipedia scale: more than 67 million articles across 300 languages, with nearly 250,000 volunteers contributing each month.