AI companies accused of hoarding and destroying millions of books

Key Points:

• Over a dozen public interest and consumer advocacy groups have called on the Federal Trade Commission (FTC) to investigate major AI developers for allegedly buying, scanning, and destroying books used to train large language models, describing the practice as "destructive" and anticompetitive.
• The groups claim AI companies purchase books in bulk, digitize their content for AI training, and then destroy the physical copies, sometimes eliminating the last surviving copies of original works.
• Critics argue this "hoard-and-destroy" practice restricts public and competitor access to essential source material and may violate Section 5 of the FTC Act, which prohibits unfair or deceptive business practices.
• The advocacy groups urge the FTC to assess the extent of the practice and determine how many destroyed books are irreplaceable originals, warning that it could concentrate control of written works among wealthy AI firms.
• The letter references a 2024 copyright lawsuit against AI developer Anthropic, accused of similar practices, though a 2025 ruling found Anthropic's use of legally purchased books for training did not infringe copyright; Anthropic has not yet commented.