Step by Step
T
Training data — is it fair use?
A central open legal question is whether training AI models on copyrighted text and images qualifies as fair use, with active, unresolved lawsuits including NYT vs. OpenAI, Getty vs. Stability AI, and major record labels vs. Suno and Udio.
Example: the New York Times suing OpenAI over allegations that ChatGPT was trained on, and can reproduce close variations of, the Times' copyrighted articles.
O
Ownership of AI-generated content — human authorship required
The US Copyright Office currently requires human authorship for copyright protection, meaning purely AI-generated output, with no meaningful human creative contribution, cannot be copyrighted.
Example: an image generated entirely by an AI model with a simple text prompt and no further human creative modification potentially being ineligible for copyright protection under current US Copyright Office guidance.
L
Licensing deals — a parallel commercial response
Alongside the ongoing litigation, some AI companies have pursued licensing deals with content creators and publishers, such as OpenAI's deals with the Associated Press and Axel Springer.
Example: OpenAI licensing news content directly from the Associated Press, rather than relying solely on the contested fair-use argument for training data.
C
C2PA watermarking — an industry response
The C2PA (Coalition for Content Provenance and Authenticity) watermarking standard has emerged as an industry response, aiming to help establish and verify the provenance of digital content, including AI-generated media.
Example: an image carrying C2PA metadata that verifies whether it was AI-generated and traces its editing history.
Applied Walkthrough
1
A major newspaper discovers that a popular chatbot can, under certain prompting, reproduce close paraphrases or even verbatim excerpts of its copyrighted articles.
2
The newspaper sues the AI company, arguing that training on their copyrighted articles without permission, and being able to reproduce similar content, violates their copyright.
3
This is precisely the kind of active, unresolved legal question represented by real cases like NYT vs. OpenAI, Getty vs. Stability AI, and record labels vs. Suno and Udio — courts have not yet definitively settled whether such training constitutes fair use.
4
Separately, some AI companies have chosen to negotiate direct licensing deals (like OpenAI's deals with AP and Axel Springer) rather than relying entirely on the contested fair-use legal argument.
Exam Application
Exams test whether you know these specific active lawsuits (NYT vs. OpenAI, Getty vs. Stability AI, record labels vs. Suno/Udio) as illustrations of the unresolved fair-use question, and whether you understand the US Copyright Office's current human-authorship requirement for copyright protection of AI-generated content.
⚠ Common Trap
The most common trap is assuming these legal questions have already been definitively resolved. As of this material, these are active, ongoing legal battles without a final settled precedent — presenting them as already legally resolved would be inaccurate.
✓ Quick Self-Check
1. What is the central open legal question regarding AI training data?
Whether training AI models on copyrighted text and images qualifies as fair use.
Tap to reveal / hide
2. Name one active lawsuit related to AI training data and copyright.
NYT vs. OpenAI, Getty vs. Stability AI, or record labels vs. Suno and Udio (any one).
Tap to reveal / hide
3. What does the US Copyright Office require for a work to receive copyright protection?
Human authorship — purely AI-generated output without meaningful human creative contribution cannot be copyrighted.
Tap to reveal / hide
4. Name one example of an AI company pursuing licensing deals rather than relying purely on fair use.
OpenAI's licensing deals with the Associated Press or Axel Springer.
Tap to reveal / hide
5. What is C2PA, and what problem is it trying to address?
A watermarking/content provenance standard, aiming to help verify the origin and authenticity of digital content, including AI-generated media.
Tap to reveal / hide