The Wikimedia Foundation filed an amicus brief in lawsuits involving The New York Times, OpenAI, and Microsoft, arguing that copyright rulings concerning AI must distinguish between training models and reproducing content for users. The organization stated that an overly simplified application of the fair use doctrine could endanger the open web, while an outright victory for AI companies could undermine licensing terms. Wikimedia noted that its websites are a critical data source for artificial intelligence and machine learning training, but also argued that services like ChatGPT reproduce text licensed freely under Creative Commons without adhering to those terms.
The Foundation's general council, Stephen LaPorte, stated, “A ruling that does not apply fair use principles carefully could threaten the sustainability of free knowledge, and stifle the development of open source technologies, ultimately impacting access to information for everyone in both cases.” Wikimedia proposed that scraping websites should not constitute copyright infringement, and that using copyrighted works to pre-train LLMs without a license should qualify as fair use if the resulting model provides “something new and different from the original sources.” However, the Foundation argued that if models reproduce material in response to user queries, thereby supplanting readers’ engagement with the original works, a license should be required.
The New York Times has argued for a more expansive definition of fair use as it applies to LLMs, alleging that OpenAI infringed copyright by scraping its website for training datasets. OpenAI and Microsoft both contend that using LLMs constitutes fair use; OpenAI stated that regurgitating work “is a rare bug,” while Microsoft reported that expert analysis found Copilot rarely reproduces full sentences from news content used to ground the model, which it said, “hardly undermines the transformative purpose of LLM training.”
