Wikimedia Foundation warns of AI copyright legal trap

The organization argues that the current legal arguments from both sides rely on oversimplified applications of the fair use doctrine. A total victory for publishers like The New York Times could allow copyright holders to gatekeep even transformative uses of information. Conversely, a sweeping win for OpenAI and Microsoft might encourage companies to ignore licensing terms, effectively fragmenting the flow of knowledge between human creators and their audiences.

This tension is existential for the Wikimedia ecosystem. While Wikipedia serves as a primary data source for training large language models, those same AI systems often siphon traffic away from the sites, undermining the volunteer communities that maintain them. The foundation also contends that services like ChatGPT frequently violate Creative Commons licenses by reproducing text without adhering to specific attribution requirements. Despite these risks, the group maintains that AI remains a vital asset for internal operations, including the detection of vandalism, doxxing, and systemic copyright infringement.

General counsel Stephen LaPorte emphasized that failure to apply fair use principles with precision threatens the future of open-source technology. Without a nuanced legal framework, the foundation fears the sustainability of free information access will be compromised, stifling both innovation and the collaborative spirit that powers the open web.