Anthropic détruit en toute légalité les livres après les avoir scannés pour alimenter son IA
2026-07-31
Yahoo actualités
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
Anthropic, developer of the Claude AI, acquired and destroyed millions of books—including rare works—to scan and train its models under 'Project Panama.' This practice led to a $1.5 billion copyright infringement settlement and raised concerns over intellectual property violations and loss of cultural heritage in the United States.[AI generated]
Why's our monitor labelling this an incident or hazard?
The event involves AI systems (large language models) being trained on massive datasets derived from physical books. The process of acquiring and destroying millions of books to create training data directly leads to harm: violation of intellectual property rights and destruction of cultural heritage. The article explicitly references a legal case confirming unauthorized use of copyrighted works, which is a breach of intellectual property rights. The destruction of books also harms communities by erasing cultural artifacts. These harms are directly linked to the AI system's development and use. Hence, the event meets the criteria for an AI Incident.[AI generated]