Jan 17 2025 59 mins 3
As revelations about Meta's use of pirated books for AI training send shockwaves through the tech industry, the battle over copyright and AI reaches a critical juncture. In this final episode of The Dynamist's series on AI and copyright, Evan is joined by FAI's Senior Fellow Tim Hwang and Tech Policy Manager Joshua Levine to discuss how these legal battles could determine whether world-leading AI development happens in Silicon Valley or Shenzhen.
The conversation examines the implications of Meta's recently exposed use of Library Genesis - a shadow library of pirated books - to train its LLaMA models, highlighting the desperate measures even tech giants will take to source training data. This scandal crystallizes a core tension: U.S. companies face mounting copyright challenges while Chinese competitors can freely use these same materials without fear of legal repercussions. The discussion delves into potential policy solutions, from expanding fair use doctrine to creating new statutory licensing frameworks, that could help American AI development remain competitive while respecting creator rights.
Drawing on historical parallels from past technological disruptions like Napster and Google Books, the guests explore how market-based solutions and policy innovation could help resolve these conflicts. As courts weigh major decisions in cases involving OpenAI, Anthropic, and others in 2024, the episode frames copyright not just as a domestic policy issue, but as a key factor in national technological competitiveness. What's at stake isn't just compensation for creators, but whether IP disputes could cede AI leadership to nations with fewer or no constraints on training data.
