In an amicus brief filed Friday, the Wikimedia Foundation urged courts hearing publishers’ cases against OpenAI and Microsoft to treat AI training on copyrighted works as fair use, while denying protection when chatbots reproduce those works for users.