What Happened?
If you own a copyright, courts have not settled whether artificial intelligence (AI) companies may train their systems on your work without permission, and Congress has not set a rule. The Congressional Research Service (CRS), Congress's in-house research arm, covers the split in its report, Generative Artificial Intelligence and Copyright Law.
Courts have gone different ways, according to a Legis1 review of the report:
- Two fair-use wins: In June 2025, two federal judges in the Northern District of California found that the AI training before them was fair use, in Bartz v. Anthropic and Kadrey v. Meta Platforms.
- One fair-use loss: On Sept. 29, a federal appeals court ruled that ROSS Intelligence's copying of Westlaw case summaries was not fair use.
- Human authorship: The D.C. Circuit held that copyright requires a human author, and the Supreme Court declined review March 2.
Why Does it Matter to Me?
If you write, draw or publish, these rulings help decide whether AI can train on your work without a license. A survey the CRS cites counts more than 150 copyright lawsuits against AI companies in the U.S.
Anthropic agreed to pay $1.5 billion to settle claims over its use of pirated books, with final approval July 20. The U.S. Copyright Office's 2023 guidance says AI-generated material alone is not eligible for protection, though a work with AI parts may qualify with enough original human authorship.
Both Sides, Now
Supporters of new protections say Congress should not leave these questions to the courts, and want stronger remedies or transparency about training data. They point to the Third Circuit's ROSS ruling, which found harm to the market for licensing Westlaw's content, and The New York Times' December 2023 lawsuit against OpenAI and Microsoft, reported by Dunya News.
The Trump administration favors letting courts decide. Its National Policy Framework for Artificial Intelligence, released March 20, says training on copyrighted material does not violate copyright law. It recommends that Congress avoid legislation interfering with the courts. In September, the Department of Justice filed a brief backing OpenAI in a copyright case.
What Happens Next?
Only Congress can write new rules. Five bills sit in committee, and each would need to clear its committee, pass the House and Senate, and win the president's signature:
- Media bill: The Content Origin Protection and Integrity from Edited and Deepfaked Media Act of 2025 would create civil liability for certain uses of copyrighted work in AI training.
- Accountability bill: The AI Accountability and Personal Data Protection Act would add civil liability involving AI and protected information.
- Subpoena bills: The TRAIN Act and its House companion would let owners seek subpoenas to learn if their works trained AI.
- Disclosure bill: The CLEAR Act would require disclosures to the Copyright Office about works used to train AI.
In the courts, the plaintiff and the Copyright Office have filed competing motions for summary judgment in *Allen v. Perlmutter*, a case on AI-assisted authorship.
---
AllCiv makes it easy to stay politically informed and involved.
---
Spot something wrong? Report an issue with this article