Anthropic’s covert acquisition and destruction of millions of books to train its AI raises copyright, legal, and supply‑chain concerns for the AI industry
Executive summary: Anthropic secretly purchased and destroyed millions of books globally to create training data for its AI, as shown in leaked judicial files describing “Project Panama.” The tactic risks copyright violations, invites regulator action, and highlights the escalating cost and legal complexity of sourcing high‑quality text for large language models.
Who is involved: Anthropic, global book sellers and publishers, U.S. and European copyright authorities, AI training data suppliers
Likely next: Regulators in the EU and US may open investigations into AI training data practices., Anthropic could be required to disclose its data‑sourcing methods in forthcoming transparency reports., Market pressure may drive AI firms toward licensed data pools or synthetic data generation.
Court documents revealed that Anthropic ran a secret operation called “Project Panama,” buying and destroying vast numbers of books worldwide to obtain training data for its models. The practice exposes the company to potential copyright infringement claims and underscores growing tension between AI firms’ data hunger and intellectual‑property rights. While the move could improve model performance, it also invites regulatory scrutiny and may push the industry toward licensed data markets or alternative training approaches.
Timeline
- — «Nous ne voulons pas que ça se sache» : Anthropic a détruit dans l’ombre des millions de livres pour alimenter son IA (Le Figaro — Économie)
- — Claude Mythos peut-il détruire l’économie mondiale et les communications ? L’IA d’Anthropic a compromis les mécanismes secrets qui sécurisent Internet (Le Figaro — Économie)
Analysis — what this means
Sectors affected
- AI model training
- Book publishing
- Semiconductor memory manufacturing
Historical parallels
- Anthropic’s Claude Mythos model broke two encryption systems, compromising internet security mechanisms (Le Figaro, July 28 2026)
- Anthropic and OpenAI executives urged the U.S. government to delay release of advanced AI models (Le Figaro, July 29 2026)
- Microsoft disclosed a $3.2 billion investment in Anthropic (TechCrunch, July 29 2026)
Key entities
Sources
- «Nous ne voulons pas que ça se sache» : Anthropic a détruit dans l’ombre des millions de livres pour alimenter son IA — Le Figaro — Économie
- Claude Mythos peut-il détruire l’économie mondiale et les communications ? L’IA d’Anthropic a compromis les mécanismes secrets qui sécurisent Internet — Le Figaro — Économie
Related cases
- AMD’s up‑to‑$5 billion stake in Anthropic signals a major push into foundation‑model AI as the startup readies its IPO
- Anthropic postpones its IPO until shortly before the November US Congressional elections, targeting a potential $2 trillion valuation
- Anthropic postpones its planned IPO, aiming for a debut before the November US congressional elections with a potential valuation of up to $2 trillion
- Sony and Warner Music file a billion‑dollar lawsuit against Anthropic alleging mass theft of copyrighted songs to train its AI models
- Federal judge rules Trump administration’s blacklist of AI firm Anthropic unconstitutional, restoring its First Amendment rights
- A US judge ruled that the Trump administration illegally retaliated against AI firm Anthropic over its Pentagon AI disputes