US Copyright Office Weighs In on AI and Copyright: Key Policy Documents
The U.S. Copyright Office has issued a series of policy reports and guidance documents addressing whether AI-generated works can be copyrighted, how training data interacts with existing copyrights, and what disclosure obligations AI users face.
The AI Initiative
The Copyright Office launched its Copyright and Artificial Intelligence initiative in 2023 to examine the intersection of copyright law and generative AI. The initiative has produced multiple reports addressing different aspects of AI and copyright law.
Key Positions
1. Copyrightability of AI-Generated Works
The Copyright Office has consistently held that works generated entirely by AI systems — without meaningful human authorship — are not copyrightable. This follows the Supreme Court’s holding in cases requiring human authorship. However, works that include human-authored elements combined with AI assistance may be registrable, provided the human-authored elements are clearly identified and the AI contributions are disclaimed.
2. Training Data and Fair Use
The Office has acknowledged that the question of whether training AI models on copyrighted works constitutes fair use is being actively litigated. Multiple lawsuits are working through the courts:
- The New York Times v. OpenAI/Microsoft (S.D.N.Y., filed December 2023) — alleges systematic copying of NYT articles for training
- Authors Guild v. OpenAI (S.D.N.Y.) — authors allege their books were used without permission to train GPT models
- Getty Images v. Stability AI (D.Del. and UK) — alleges unauthorized use of millions of copyrighted images to train Stable Diffusion
3. AI Training Data Transparency — California AB 2013
At the state level, California passed AB 2013 (signed September 2024), which requires generative AI providers to publicly post documentation about the data used to train their models, including:
- Sources of the training data
- Whether the data includes copyrighted works
- Whether the data includes personal information
- A description of how the data was collected and processed
This law takes effect January 1, 2026, and applies to generative AI systems released or substantially modified on or after January 2, 2025.
Disclosure Requirements
The Copyright Office has recommended that Congress enact legislation requiring disclosure of AI-generated content, particularly in contexts where the public has a right to know whether content is human-authored. This aligns with broader AI disclosure trends seen in:
- The EU AI Act, which requires providers of AI systems that generate synthetic media to disclose that content is AI-generated (Article 50)
- California’s deepfake laws (AB 2655, AB 730)
Primary Sources
- Copyright and Artificial Intelligence — U.S. Copyright Office
- California AB 2013 Full Text — California Legislative Information
- EU AI Act (Regulation 2024/1689), Article 50 — EUR-Lex
- California AB 2655 (Deepfakes) — California Legislative Information
AI automated and generated content. Human reviewed. All policy text is summarized from the primary sources linked above.
The Weekly AI Policy Brief
AI policy moves fast. Get the week’s developments, new trackers, and podcast episodes in one short email — free, ad-free, no spam.
Leave a Reply