Reuters reports courts leaning against AI training fair use claims

U.S. courts are increasingly signaling that unauthorized AI model training may not qualify as fair use, according to a Reuters analysis of recent litigation trends. Multiple federal judges have expressed skepticism toward tech companies arguing that ingesting copyrighted works for machine learning constitutes transformative use. This judicial shift challenges the foundational legal defense employed by major AI developers facing class-action lawsuits from authors, artists, and publishers. Legal experts warn that adverse rulings could establish binding precedents requiring licensing agreements for training data. The evolving case law suggests plaintiffs have gained momentum in arguments that AI models compete directly with original creators. Industry representatives maintain that training represents non-expressive technical processing distinct from copyright infringement. However, preliminary injunction hearings and oral arguments indicate judges are weighing market harm factors more heavily than previously anticipated. These developments create significant uncertainty for AI companies relying on unlicensed internet-scraped datasets for model development.