The US Copyright Office’s AI Guidance Explained

Most of what gets written about AI and copyright is commentary. The Copyright Office itself has actually published its own position, more than once, and collected all of it in one place at copyright.gov/ai. If you want to see the government’s actual reasoning rather than someone’s summary of it, that page is worth bookmarking.

It starts with registration guidance from 2023, which addresses a practical question: what happens when you try to register a work that includes AI generated material. The short answer is that you have to disclose which parts were generated by AI, and only the parts you actually authored, the human creative choices you made in selecting, arranging, or modifying that output, are what get protected.

The Copyright Office then followed with a three part report. Part 1, released in 2024, focused on digital replicas: AI generated recreations of a real person’s voice or likeness, and recommended that Congress pass a federal law addressing them, since existing law does not cover the issue well. Part 2, released in early 2025, tackled the question at the center of most AI ownership disputes: whether AI generated output can be copyrighted at all. The Office’s position held steady with what courts have also found, that purely AI generated content with no meaningful human creative input is not eligible for protection, while output a human meaningfully directs, edits, or arranges can be. Part 3 addressed a different question entirely: whether training an AI model on copyrighted material counts as fair use, and concluded that there is no blanket answer, it depends on the specific use.

None of this replaces legal advice for your specific situation, but it does mean you no longer have to guess at where the government stands. If your company is building on AI, or your product touches AI generated content in any way, it is worth knowing this guidance exists before a dispute forces the question.