Legal Landscapes

Generative AI has thrown a grenade into copyright law. When a large language model is trained on millions of copyrighted texts — news articles, novels, academic papers, song lyrics — is that “fair use” or theft? The legal battles unfolding right now will shape who profits from AI and who is dispossessed by it. Reclaiming Fair Use provides the historical context, tracing how copyright doctrine has been contested and reshaped by new technologies from the photocopier to the VCR. Our guest, Demetri Blaisdell from The New York Times, brings this into the present tense — the Times’ lawsuit against OpenAI is one of the landmark cases testing whether AI companies can build empires on other people’s creative labor. Kyle Courtney’s argument that “training isn’t copying” offers the counterpoint, and the debate between these positions is far from settled.

Who owns the output of a machine trained on everyone’s work? How do legal frameworks designed for an analog world apply to systems that can generate infinite variations of existing content? And what happens to the creators — journalists, artists, scholars — whose work feeds the machine but who see none of its profits?