The Wikimedia Foundation has filed an amicus brief in the New York Times copyright lawsuit against OpenAI and Microsoft, advocating for a nuanced approach to fair use that protects the open web. The organization warns that an extreme decision in either direction could threaten the sustainability of free knowledge and stifle open source technology development. Wikimedia argues that a restrictive fair use ruling could lead to a closed web with fewer secondary source materials, while a blanket fair use ruling could weaken knowledge availability.
The foundation proposes applying fair use to specific AI use cases, including certain model training scenarios and retrieval-augmented generation (RAG), provided LLMs do not regurgitate pre-existing articles without attribution. Regarding model training specifically, Wikimedia suggests scraping copyrighted content should be considered fair use if used to train LLMs that are "something new and different from the original sources." The brief notes Wikipedia is considered one of the highest-quality datasets for training LLMs. General Counsel Stephen La Porte emphasized that careful application of fair use principles is essential to avoid threatening free knowledge sustainability and access to information for everyone. A summary judgment in the lawsuit is expected soon, as this case could set an important precedent among over a dozen similar copyright cases working through the courts.
Source: Engadget · Summarized by HeadlinesBriefing