Bluesky user argues AI training is fair use not theft

A Bluesky post published September 26, 2026, asserts that training artificial intelligence models on copyrighted works constitutes fair use rather than theft because the process does not deprive owners of their original property. The author, ecutruin.bsky.social, argues that legal precedents have already established AI training as permissible under copyright law, distinguishing it from traditional intellectual property violations. This statement reflects a recurring defense utilized by AI proponents to counter accusations of data misappropriation by creators and rights holders. The argument hinges on the non-rivalrous nature of machine learning ingestion, positing that copying for computational analysis differs fundamentally from market substitution. Legal experts note that while some courts have explored fair use defenses in AI litigation, no binding Supreme Court precedent currently settles whether large-scale model training qualifies as transformative use across all jurisdictions.