Wikipedia chooses the middle ground in New York Times and OpenAI lawsuit

Three years after The New York Times filed a lawsuit against OpenAI and Microsoft, Wikipedia has entered the debate on the side of Wikipedia. The Wikimedia Foundation announced today that it has filed an amicus brief urging the court to find a middle ground in the lawsuit, which centers around whether OpenAI and Microsoft infringed on copyright by training on Times articles and reproducing portions of those articles.

The Foundation believes there are instances when this should be permissible, and instances when it should not.

According to the brief, a ruling in favor of overly restrictive fair use could impede technological innovation, limiting the availability of tools that support open source projects like Wikipedia. Conversely, a ruling in favor of blanket fair use could result in a more closed web, weakening the availability of secondary sources of knowledge online.

In essence, the Foundation argues that large language models (LLMs) should be able to train on copyrighted materials, but those copyrighted materials should not be used in retrieval-augmented generation, such as in chatbots or Google's AI overview.

Wikimedia Foundation General Counsel Stephen LaPorte commented, "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." The Verge notes that Wikipedia has had a complicated relationship with AI for some time; these LLMs have trained on Wikipedia contributors' work while also diverting traffic away from the free encyclopedia, and may reproduce copyrighted material licensed under Wikimedia's Creative Commons without adhering to the relevant licenses.

Earlier this year, Wikipedia took steps to ban AI-generated content from its pages. However, Wikipedia maintains that LLMs and AI tools have assisted editors in detecting plagiarism and vandalism.