Matt Blaszczyk (University of Georgia School of Law) has posted Courts Won’t Settle the Fight Over AI and Creative Work: Politics Will (Bulletin of the Atomic Scientists, July 14, 2026) on SSRN. Here is the abstract:
Currently, more than 100 lawsuits are making their way through United States courts, testing whether AI companies may use copyrighted works to train their models without permission or payment to copyright owners. The first two decisions, Bartz v. Anthropic and Kadrey v. Meta, both from a single federal district court in California, suggest that the training itself is likely fair use, though the two judges split over the use of books pirated from shadow or unauthorized libraries and over whether copyright should evolve to shield authors from competing with machines trained on their own works. Neither ruling binds other courts, and the harder questions—who and what exactly infringes, and what a courtroom victory would deliver for creators—remain open. Litigation will not yield bright-line answers anytime soon, and courts are unlikely to deem AI training simply “theft.” This uncertainty serves the biggest players on both the content and AI sides, while most individual creators cannot count on copyright to ensure they are compensated for AI training that aims to replace them. One way out of the legal quagmire would lie in political action: the introduction of a licensing regime that could ensure compensation for actual creators, an opt-out mechanism for those unwilling to settle for a statutory fee, and a broader democratic debate concerning the power imbalances that both technology and media impose on the people.
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum