The UNC Journal of Law and Technology (JOLT) just published John Kivus’ detailed article Generative AI and Copyright Law: A Misalignment That Could Lead to the Privatization Of Copyright Enforcement. The article is a well-rendered deep dive into US copyright law, its recent history in the courts, and the complex implications of how generative AI can affect IP owners, businesses, and the courts. For software developers and rights holders, the article presents an intuitive look at AI generative concepts, such as how generative AI creates new data through the use of large language models (LLMs), word vectors, and word embedding and transformer models. It also examines how generative AI models are trained.
John then lays out the many difficulties courts and IP owners may encounter applying existing copyright law to generative AI. He walks through the recent history of generative AI lawsuits focused on training AI models and how even more recent cases have centered on claims related to outputs. In this sharply articulated work, John not only leans into case law germane to AI and copyright but examines current USPTO rulings and how these rulings may transition the protection of copyrights into the world of private enforcement. The article is a must-read for anyone in the AI space.
Click HERE to download a copy.
