Wikimedia Foundation files amicus brief in landmark AI copyright case

The Verge reports that the Wikimedia Foundation, the nonprofit behind Wikipedia, filed an amicus brief on a Friday in a set of AI copyright lawsuits where publishers such as The New York Times are suing OpenAI and Microsoft. Wikimedia supports neither side. It argues that both parties are pushing overly broad and simplified readings of fair use, and warns that a clear win for either could harm free knowledge.

According to the article, Wikimedia fears that a complete victory for publishers would let rights holders gatekeep even innovative, transformative uses of their work, while a victory for AI companies would amount to permission to ignore licensing terms. The foundation says its sites are among the most important sources of AI training data, yet AI systems draw visitors away from the volunteer communities that maintain them. It also contends that ChatGPT-style services reproduce Creative Commons text without honoring its license. At the same time, it notes that language models help editors counter vandalism, spot copyright violations and doxxing, and find articles needing work.

The proposed compromise has two parts. Scraping websites should not count as infringement, and pre-training models on copyrighted works without a license should be fair use when the result offers something new and different from the sources. But when models retrieve copies of material in answer to queries and deliver it in ways that replace readers' engagement with the originals, a license should be required, since the output merely substitutes for the work. For Wikipedia, that would mean credit under its Creative Commons license.

The Times argues for a broader view of fair use, while OpenAI and Microsoft say model use is fair use. Wikimedia's CEO said the foundation took a position to defend open human knowledge.