Скачать или смотреть AI Copyright vs Fair Use, Explained: Are LLMs Trained on Stolen Work?

What happens when generative AI learns by absorbing the work of writers, publishers, and artists without asking, crediting, or paying? The debate over AI copyright, AI fair use, and AI training data is no longer theoretical. It’s becoming a defining battle over copyright and AI, the future of creative markets, and whether training on copyrighted work is “fair use” or copyright infringement AI.

In this episode of THE PEOPLE'S AI PODCAST, presented by The Vana Foundation (‪@vanahq‬), we examine the growing wave of generative AI lawsuits and the core question at the center of the storm: is training AI on books fair use, or is it AI piracy at industrial scale. We explore why some legal arguments treat training as potentially permissible, while the bigger flashpoint may become the second bucket of cases: when AI tools generate outputs that are too close to protected expression, from copyrighted characters to text that looks like AI plagiarism.

We hear from children’s and YA author Rachel Vail on what it feels like to discover her books were used to train models, and why “consent” matters as much as compensation. Stanford law professor Mark Lemley breaks down AI and intellectual property, including why the law may distinguish training legality from infringing outputs. Medium CEO Tony Stubblebine lays out a practical framework for creators and platforms: consent, credit, compensation, and how power dynamics shape this fight.

We also zoom out to the web’s counteroffensive: efforts to block AI crawlers, debates over robots.txt AI training, and emerging models like pay per crawl, as platforms experiment with how to protect creators while preserving public access.

Presented by Vana Foundation.

Vana supports a new internet rooted in data sovereignty and user ownership—so individuals (not corporations) can govern their data and share in the value it creates. Learn more at vana.org.