Wikimedia urges courts to distinguish AI training from copying in copyright cases

The Wikimedia Foundation filed an amicus brief Friday urging courts to protect transformative AI training as fair use while requiring licenses for outputs that substitute for original works, The Verge reported. The filing addresses publisher lawsuits against OpenAI and Microsoft, including The New York Times' case. Wikimedia argues that language models can help Wikipedia editors detect vandalism, copyright violations and doxxing. But it says services that reproduce Wikipedia text without following its Creative Commons terms violate copyright and draw visitors away from the communities maintaining the site. Those terms require attribution and require adaptations to retain the same license.