The essentials
Anthropic allegedly sourced training data through piracy, torrenting, and unauthorized web scraping.
New allegations detail unauthorized methods Anthropic used to acquire training data for its AI systems. According to claims, the company obtained pirated books and torrented content to build its datasets. Additionally, Anthropic employed web scraping to harvest copyrighted works without authorization.
These allegations emerge as part of the broader copyright infringement litigation against the AI company. The specific mechanisms of data acquisition represent a distinct dimension of the copyright disputes, moving beyond reproduction claims to focus on how training materials were originally obtained. The revelations underscore questions about data sourcing practices in AI development.
Other ways to read this story.
Alternative framings generated by SigNews. These are not independent sources or attributed quotations.
Unauthorized acquisition methods raise questions about AI training standards
Anthropic allegedly used pirated books, torrenting, and web scraping to gather training data. These specific acquisition mechanisms suggest systematic approaches to obtaining copyrighted content without permission. The allegations point to deliberate methodologies rather than incidental reproduction, potentially indicating knowledge of unauthorized sourcing.
Copyright disputes expand to include data sourcing alongside content reproduction claims
The allegations introduce a new layer to existing lawsuits, moving beyond whether AI systems reproduced copyrighted works to how training datasets were originally compiled. Claims of piracy, torrenting, and web scraping represent distinct legal questions about acquisition versus use, broadening the scope of copyright infringement arguments against the company.