Wikimedia Urges Courts to Protect AI Training Without Shielding Chatbots From Copyright Liability
The Wikimedia Foundation has filed an amicus brief in a group of copyright lawsuits involving publishers such as The New York Times, OpenAI, and Microsoft. The organization is not siding fully with either side, instead proposing a distinction between training language models on copyrighted material and deploying chatbots that reproduce that material in their responses.
Wikimedia argues that courts should protect the former because it supports the development of AI-based tools used by Wikipedians. At the same time, it says that chatbots should remain accountable when they regurgitate copyrighted work to users.
The foundation warns that a ruling that does not recognize this middle ground could have consequences for the open web. A broad prohibition on training with copyrighted material could deprive Wikipedia contributors of useful tools, while a ruling that treats the technology as broadly exempt from copyright rules could undermine the ability to share knowledge online.
The cases are among a set of landmark disputes over how copyright law applies to generative AI. Wikimedia's position emphasizes the need to preserve both access to AI development and respect for publishers' rights.