In the US, the Copyright Office is seeking comments on how to address the issue, with policy proposals expected by summer 2024. Meanwhile, Japan initially seemed to favor AI developers, but recent developments suggest a shift towards protecting creators' interests. The EU's AI Act requires transparency reports on all data used to train AI models, regardless of copyright status. The article also mentions the New York Times' lawsuit against AI companies for unauthorized use of its content, which could set a precedent for future cases.
Key takeaways:
- The use of copyrighted works as training data for AI models is a contentious issue, with unresolved questions about its legality and the potential for regulation, legislation, litigation, or commercial agreements to address it.
- Japan has been seen as favorable to AI learning due to its copyright law allowing the use of copyrighted works for data analysis, but this may change due to concerns about the interests of creators.
- The European Union's AI Act requires providers of general-purpose AI models to publish transparency reports about the content used for training, allowing copyright holders to opt out of their work being data mined.
- In the US, the Copyright Office, Congress, and the courts are all involved in efforts to address the issue, with potential for legislative reforms, transparency requirements, and legal rulings to influence the use of copyrighted works in AI training.