What happened
More than 100 authors filed a lawsuit against Anthropic in Northern California on June 17, 2026, demanding over $75 million for alleged copyright infringement. The complaint states Anthropic used over 500 pirated books, including works by Tiffany Aliche and Laura Esquivel, to train its AI systems. Anthropic allegedly acquired these via BitTorrent from illegal websites, with the lawsuit specifically alleging redistribution. This action follows a September 2025 class-action settlement where Anthropic paid $1.5 billion for similar claims; these plaintiffs opted out.
Why it matters
This lawsuit, specifically alleging redistribution of copyrighted material via BitTorrent, escalates legal scrutiny on AI training data provenance beyond initial acquisition. For procurement teams and security architects, this highlights the critical need for auditable data acquisition and retention records for all AI training datasets. Copyright risk around model training is an evidence-management problem, requiring provable licences, chain-of-custody logs, and deletion controls. Plaintiffs opting out of a $1.5 billion class-action settlement signals increasing legal challenges for AI developers regarding intellectual property.
