Fair Use Benefits Everyone

Commenters largely viewed OpenAI’s defense of fair use as self-serving, pointing to its commercial use of creators’ work, closed models, and objections to competitors distilling those models. Several argued that the company’s rhetoric about benefiting everyone clashes with privatized profits, and that training on humanity’s shared output should bring shared ownership or compensation. Others framed the position as corporate hypocrisy: a company claiming broad rights to use others’ work while resisting similar uses of its own. One commenter also noted that fair-use rules differ by jurisdiction, including the EU.

The legal disagreement centered on whether AI training is transformative and whether a model’s outputs substitute for copyrighted works. Supporters of broad fair use said training on material differs from reproducing it, and that distilling a model into a similar but distinct one can also qualify. Critics argued that commercial, large-scale harvesting is not automatically fair use, especially when models can reproduce articles or compete with their markets; one compared it to publishing a near-copy of a book after reading it. Commenters also distinguished training from other conduct: Simon Willison said the New York Times case includes memorization and search-and-summary features, while Anthropic’s settlement concerned pirated books despite a separate fair-use ruling on training with purchased scans. Others questioned whether a business built on automating fair use can itself claim the doctrine.