Summary
A United States judge ruled that using books to train artificial intelligence (AI) software does not break US copyright laws if the use is transformative, meaning it changes the work in a significant way. However, the judge stated that the AI firm Anthropic must face trial for keeping pirated copies of books, which could lead to fines. This case is one of the first of its kind and could influence future legal decisions about AI learning from existing content.
Key Facts
- A judge decided that AI company Anthropic's use of books to train its software is allowed under US law as transformative use.
- Three authors sued Anthropic, claiming their books were used to train Anthropic's AI without permission.
- The judge rejected Anthropic's move to dismiss the lawsuit, meaning a trial will happen.
- Anthropic is accused of storing over seven million pirated books, which could result in fines up to $150,000 per work.
- The case focuses on whether AI can legally use existing material, a topic involved in many legal disputes.
- The judge noted that the authors did not allege their works were copied directly in the AI's outputs.
- Other companies, like Disney, have filed similar lawsuits against AI firms for using copyrighted content.
- Anthropic's defense of "fair use" was accepted, but the company still faces a trial regarding how they obtained the books.
