On April 13, 2024, internal documents leaked from Anthropic revealed the existence of a clandestine initiative dubbed 'Project Panama'. The core objective of this project is to amass a global repository of book content by employing destructive scanning techniques. This approach ensures the acquisition of pristine training datasets, untainted by AI-generated material. Anthropic covertly procures, in large quantities and through intermediaries, paper books published prior to 2022. These books undergo a process where their spines are surgically removed using hydraulic cutters. Following high-speed scanning, the physical copies are then shredded and disposed of. Estimates suggest that the project has already processed millions of volumes, with financial outlays reaching into the tens of millions of dollars. Despite a U.S. Federal Court ruling that classified this practice as fair use, the initiative has ignited a firestorm of debate concerning the preservation of cultural heritage, data ethics, and the delineation of legal boundaries.
