Copyright

Legal protection granting creators exclusive rights to their original works, now central to debates over AI training data and outputs.

Copyright grants creators exclusive rights to reproduce, distribute, and build upon their original works—a framework now colliding with AI.

AI Challenges

Training data: Can AI legally train on copyrighted works?

• Fair use argument: Transformative learning, not copying
• Creators’ view: Unauthorized use of their work
• Ongoing lawsuits: NYT v. OpenAI, Getty v. Stability AI

AI outputs: Who owns AI-generated content?

• Human authorship traditionally required
• US Copyright Office: “Purely AI works not copyrightable”
• Human-AI collaboration: Uncertain territory

Current State

• No clear legal consensus globally
• EU AI Act addresses transparency, not copyright
• Opt-out mechanisms emerging (robots.txt, Do Not Train)
• Licensing deals between AI companies and publishers

Implications

The resolution will shape who benefits from AI—creators, platforms, or AI developers—and whether training data requires compensation.