Anthropic Book Orders Fuel Suspicions of AI Data Sourcing
TL;DR. Independent booksellers in Europe report unusual bulk orders of obscure titles, fueling suspicions that AI companies are quietly acquiring data for chatbot training. - Anthropic's Project Panama previously scanned millions of books to train its Claude AI models. - Large, eclectic orders with no negotiation mirror methods used to build AI datasets. - German publishers consider mass scanning practices illegal under local copyright law.
- Booksellers in Europe are receiving suspicious bulk orders for niche books.
- Similar activity in the US was linked to Anthropic's 'Project Panama' for AI training data.
- Anthropic reportedly spent 'tens of millions of dollars' to scan books for Claude.
- German publishing associations deem such scanning illegal under local law.
- The practice highlights legal and ethical questions surrounding AI model training data acquisition.
Sources
- A 'bananas' order for 5000 obscure book titles fuels suspicion — irishtimes.com
- techxplore.com — techxplore.com