Anthropic's $1.5 Billion Copyright Settlement — AI Training Ruled Legal, FSF Demands Open AI Models

Anthropic Copyright Lawsuit — Why the Court Ruled AI Training as Fair Use

The lawsuit, known as Bartz v. Anthropic, was a class action filed by fiction and nonfiction authors against Anthropic. The core issue was simple: did Anthropic infringe copyright by downloading books from Library Genesis (aka Libgen, a site that freely shares books online) and Pirate Library Mirror to use for AI training?

In June 2025, a U.S. federal judge ruled in Anthropic's favor: “Using books for AI training constitutes Fair Use.” However, one issue remained unresolved — whether the act of downloading books from illegal sites was itself legal. Ultimately, both sides chose to settle for $1.5 billion rather than go to trial.

FSF Rejects $1.5 Billion Settlement — Why They Demand Open AI Models

The most striking aspect of this lawsuit was the stance of the Free Software Foundation (FSF). Founded by Richard Stallman, the FSF has upheld the philosophy that ‘all software should be free’ for over 40 years.

It was confirmed that the FSF's book «Free as in Freedom: Richard Stallman's Crusade for Free Software» (by Sam Williams) was included in Anthropic's training data. This book was published under the GNU Free Documentation License (a license that allows anyone to freely use the work).

Normally, one would accept the settlement money, but the FSF refused the payment. Instead, they made this demand:

In short: “Don't give us money — open up the AI.” Currently, AI systems like ChatGPT and Claude disclose nothing about what data they were trained on or how they work internally. The FSF argues this is a contradiction — taking freely available internet knowledge while keeping the results proprietary. Understanding how AI learns and operates helps grasp the heart of this debate.

AI Copyright Lawsuits — Anthropic and OpenAI Face Billions in Claims

Anthropic isn't the only one being sued. In January 2026, music publishers including Universal Music Group (UMG) filed an additional $3 billion lawsuit against Anthropic, claiming the company used 20,000 copyrighted songs for training without authorization.

Similar lawsuits are spreading across the industry. Encyclopaedia Britannica has already sued OpenAI for 100,000 counts of copyright infringement. The total cost the AI industry may have to pay for training data issues is expected to exceed billions of dollars.

Copyright Issues Creators and Professionals Need to Know in the AI Era

This case affects everyone — those who build AI, those who use AI, and the original creators whose content AI was trained on.

If you're a creator — a writer or artist — you should be aware that your work may have been used for AI training. You can check the Anthropic Copyright Settlement site to see if your works were included and whether you're eligible to file a claim.

If you use AI at work — the origin of AI training data is becoming an increasingly important issue. Especially in corporate environments adopting AI, the question ‘What data was this AI trained on?’ could become central to assessing legal risk.

Whether the day will come when AI models and training data are fully disclosed, as the FSF demands, remains uncertain. But one thing is clear: In the AI era, the question ‘Whose knowledge was used to build this AI?’ will echo louder and more frequently.

If you want to learn more about AI and vibe coding, check out our Free Learning Guide.