Anthropic clears $1.5 billion copyright settlement over pirated AI training data
A US federal judge has approved a record $1.5 billion payout to authors over pirated books, yet the underlying legality of using copyrighted text to train artificial intelligence remains unresolved for the global tech sector.
Judge Araceli Martinez-Olguin signed off on Monday on the final approval of Anthropic’s $1.5 billion class-action settlement. The payout resolves claims that the artificial intelligence lab illegally acquired millions of copyrighted books to build its foundational models.
The record sum is believed to be the largest in United States copyright history. It will distribute $3,000 per work, with compensation shared among the authors and publishers holding rights to an estimated 500,000 affected titles.
The financial penalty stems directly from how the company sourced its training data. While Anthropic legally purchased and scanned some books, it also downloaded millions of titles from pirate platforms including Library Genesis and Pirate Library Mirror.
Retired Judge William Alsup, who initially presided over the case, previously ruled that this specific method of data acquisition was strictly illegal. However, Alsup simultaneously decided that the subsequent act of training an artificial intelligence model on copyrighted text constitutes fair use.
This dual ruling creates a highly complex landscape for technology investors and European enterprises navigating the artificial intelligence sector. The massive financial penalty addresses the illicit procurement of data. Meanwhile, the fair use designation currently protects the core mechanics of model development.
Many authors and creators argue this outcome falls short of a definitive victory for intellectual property rights. Crucially for the broader market, this resolution fails to establish binding legal precedent across the global technology industry.
Because Anthropic settled before an appeals court could review the fair use decision, the case will never set a universal standard. Other federal judges remain entirely free to interpret copyright law differently based on their own specific cases.
This ongoing legal uncertainty continues to cast a long shadow over major technology firms. It complicates valuation models for companies relying heavily on large language models. A string of copyright lawsuits remains active against industry giants including Google, Meta, Midjourney, and OpenAI.
The litigation is actively expanding as creators seek to protect their livelihoods in the digital age. Just last week, a coalition including Hachette, Cengage, Elsevier, author Scott Turow, and S.C.R.I.B.E. filed a new class-action lawsuit against Google.
This latest legal challenge accuses the search giant of utilizing their copyrighted materials to train its Gemini artificial intelligence platform. The outcome of these future battles will ultimately dictate the financial and operational rules for the entire European and global technology economy.
