A recent complaint filed with the US Federal Trade Commission has accused AI companies of engaging in a practice known as 'scan-and-destroy', where they buy physical books, scan them, and then destroy the originals to train their models. This practice has sparked concerns over anticompetitive behavior and the potential for monopolized knowledge.
Key Insights
Companies like Anthropic and Amazon have been accused of hoarding knowledge and preventing access to resources through this practice. Anthropic's 'Project Panama' aimed to 'destructively scan all the books in the world', with the company using a codename to keep the project secret. The advocacy groups argue that this practice prevents competing AI developers and the public from accessing resources, raising concerns over the long-term implications for innovation and knowledge sharing.
