AI Is Destroying Millions of Books (But Not For The Reason You Think)
Summary
Taylor Lorenz and copyright lawyer Derek Slater debunk viral claims that Anthropic is destroying rare books for AI training. They explain that the company scans discarded, out-of-print volumes to mitigate massive copyright liability, as retaining physical copies risks billions in statutory damages. The discussion highlights how restrictive copyright laws force this destructive digitization, urging listeners to support libraries and archives instead of blaming AI firms.
Topics discussed
Sponsors: Indeed, Toyota, Accenture, San Pellegrino
Anthropic's book scanning and viral misinformation
Clarifying the source and rarity of scanned books
Sponsors: Indeed, Toyota, Accenture, San Pellegrino
Google Books history and Anthropic's competitive drive
Creator support plea: Patreon and Substack
Legal reasons for destructive scanning: Copyright risks
Copyright damages and the impossibility of licensing
Fair use, security controls, and supporting libraries
Internet Archive lawsuits and out-of-print books
Sponsors: Indeed, Toyota, Accenture, San Pellegrino
Public domain distinctions and episode conclusion
Listen ad-free on Castria