The Vergecast episode explores the controversial use of millions of destructively scanned books to train Anthropic’s Claude AI, highlighting the legal and ethical issues around sourcing copyrighted material for AI development. It also discusses Netflix’s evolving approach to theatrical releases amid industry changes, and touches on ongoing frustrations with smart home device compatibility.
The episode of The Vergecast opens with host David Pierce discussing his ongoing quest to find a new smartphone, debating between sticking with iOS or switching to Android, and inviting listeners to share their suggestions. He mentions his dissatisfaction with his current iPhone’s battery life and his openness to experimenting with various devices, including foldables and Pixels. However, the main focus of the episode quickly shifts to two major topics: the controversial use of books in AI training and the evolving role of movie theaters in the streaming era.
The first major segment features Will Oremus from The Washington Post, who discusses his reporting on Anthropic’s “Project Panama.” This project involved the destructive scanning of hundreds of thousands—possibly millions—of books to digitize their contents for training the Claude AI model. The process included slicing the spines off books for efficient scanning, often sourcing them from massive used book warehouses. The conversation highlights how books are seen as high-quality, vetted content, making them especially valuable for AI training compared to the noisier data found online.
Oremus explains that Anthropic, like other AI companies, initially relied on pirated “shadow libraries” such as LibGen to acquire digital books, a practice also attributed to OpenAI. When attempts to license books legally proved too slow and expensive, Anthropic turned to bulk purchasing and destructive scanning. The discussion delves into the legal and ethical complexities of this approach, noting that while courts have sometimes ruled that training AI on copyrighted books can be considered fair use, the method of acquiring those books—especially through piracy or without proper licensing—remains a contentious legal battleground.
The episode then transitions to a conversation with Julia Alexander, a media analyst at Puck, about Netflix’s evolving relationship with movie theaters. As Netflix considers acquiring Warner Bros. Discovery, questions arise about whether the streaming giant is genuinely committed to theatrical releases or simply making promises to appease regulators. Alexander discusses the business logic behind theatrical releases, noting that movies with a theatrical window tend to perform better on streaming platforms due to increased awareness and perceived quality. However, she also points out the challenges facing theaters, including declining attendance, changing consumer preferences, and the need for new types of communal experiences beyond traditional film screenings.
Finally, the show addresses a listener’s frustration with IKEA’s new Matter-enabled smart home buttons not working with Google Home, highlighting broader issues with smart home interoperability and the slow adoption of new standards. The episode closes with reminders for listeners to participate in the show’s hotline, share their tech experiments, and stay tuned for future episodes covering ongoing tech news and industry developments. Throughout, the hosts emphasize the rapid pace of technological change, the ethical dilemmas it creates, and the importance of community feedback in navigating these shifts.