Under U.S. copyright law, only works created by a human being can be copyrighted, so purely AI-generated content receives no federal protection and falls into the public domain. That is the governing principle behind every current question about AI copyright rules, and a federal appeals court reaffirmed it in March 2025. The picture shifts, though, when a person uses AI as a tool inside their own creative process: the human-authored portions of that work can be registered and protected, as long as the applicant discloses the AI-generated parts and follows the Copyright Office’s rules.
Purely AI-Generated Work Has No Copyright
The Copyright Office refuses to register works “produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.”1U.S. Copyright Office. Works Containing Material Generated by Artificial Intelligence That position was tested in Thaler v. Perlmutter, in which computer scientist Stephen Thaler tried to register an image called “A Recent Entrance to Paradise” generated autonomously by his Creativity Machine, listing the machine as the sole author. The Copyright Office denied the application. In March 2025, the U.S. Court of Appeals for the D.C. Circuit affirmed, holding that “the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being.”2United States Court of Appeals for the District of Columbia Circuit. Thaler v. Perlmutter
The court was careful about what it did not decide. The ruling does not bar copyright for works made with AI assistance. Human authorship, the court said, “requires only that the author of that work be a human being — the person who created, operated, or used artificial intelligence — and not the machine itself.”2United States Court of Appeals for the District of Columbia Circuit. Thaler v. Perlmutter
The practical consequence for anyone publishing raw AI output: no legal claim over it. If you type a prompt into an image generator and post the result without meaningful creative work, anyone can copy, modify, sell, or redistribute that image without asking you.
When AI-Assisted Work Can Be Copyrighted
In March 2023, the Copyright Office published formal guidance on registering works that combine human and machine contributions.3Federal Register. Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence The rule is that copyright covers only the parts a human actually authored. The AI portions get no protection, even if they sit inside an otherwise registrable work.
Writing a detailed prompt does not count as authorship. The Copyright Office treats a prompt as an instruction, closer to telling a freelance illustrator what to paint than to painting it yourself. To earn protection you have to do more: edit the AI output substantively, combine it with your own original material, or make creative choices about how AI-generated pieces are selected, arranged, and organized into a larger composition.
A small, incidental amount of AI content in an otherwise human work may be treated as de minimis and require no formal exclusion. Once the AI-generated portion is significant, though, you need to mark clearly which contributions are yours and which belong to the machine. Registration examiners look at these applications case by case, and the question they ask is whether a human exercised genuine creative control over the final result.
How to Register an AI-Assisted Work
You register through the Electronic Copyright Office (eCO) at copyright.gov, and for any work that contains AI-generated material you have to use the Standard Application rather than the single-work short form.
The key entry is the “Author Created” field, where you describe your own contribution specifically. The Copyright Office suggests wording along the lines of “Selection, coordination, and arrangement of [human-authored content] created by the author and [AI content] generated by artificial intelligence.” If the AI-generated content is more than trivial, you also exclude it from the claim in the “Limitation of the Claim” section, using a short description under “Material Excluded” such as “illustrations generated by artificial intelligence.” Do not list an AI tool or its developer as an author or co-author.3Federal Register. Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence
The disclosure obligation is Copyright Office policy, not a duty written into the statute. Section 409 of Title 17 requires applicants to identify authors, describe preexisting material, and provide “any other information regarded by the Register of Copyrights as bearing upon the preparation or identification of the work.”4Office of the Law Revision Counsel. 17 U.S. Code 409 – Application for Copyright Registration The 2023 guidance uses that authority to require AI disclosure, and states that applicants have a duty to disclose AI-generated content. A registration obtained without that disclosure can be cancelled if the omission comes to light later.
Fees and Processing Times
The Standard Application filing fee is $65. Sole authors who are also the sole claimant for a single, non-work-for-hire piece may qualify for a $45 fee instead.5U.S. Copyright Office. Fees After payment, you upload a digital deposit copy (usually PDF or JPEG) and receive a service request number by email.
Electronic applications with digital deposits that go through without correspondence average about two months, with some running up to about four months. When the examiner needs to write back to the applicant, which happens for roughly a quarter of electronic filings, the average climbs to around four months and can reach eight or more. Paper filings take considerably longer.6U.S. Copyright Office. Registration Processing Times FAQs AI-assisted applications, because they force an examiner to sort human-authored elements from machine-generated ones, are more likely to draw follow-up questions, so plan for the longer end of the range.
Training AI on Copyrighted Work
Building a generative AI model takes enormous amounts of training data, and those datasets often include copyrighted books, images, music, and code pulled from the web. Whether this training infringes copyright is the central issue in a wave of pending lawsuits from authors, visual artists, and music publishers against AI developers.
The developers’ main defense is fair use. Section 107 of Title 17 tells courts to weigh four factors: the purpose and character of the use (including whether it is commercial), the nature of the copyrighted work, how much of the work was used, and the effect on the market for the original.7Office of the Law Revision Counsel. 17 U.S. Code 107 – Limitations on Exclusive Rights: Fair Use AI companies argue the training is transformative because models learn patterns rather than storing copies, and the output serves a different purpose than the originals. Rights holders respond that training directly displaces demand for their work and that some models can produce near-copies of training data when prompted the right way.
No appellate court has yet resolved whether training a commercial AI model on copyrighted material is fair use. In Thomson Reuters v. Ross Intelligence, a Delaware federal court rejected the fair use defense at summary judgment in a case involving an AI legal research tool trained on copyrighted headnotes, finding the first and fourth factors favored the copyright holder. Other major cases, including the New York Times’s suit against OpenAI and music publishers’ claims against various AI developers, are still working through the courts.
A separate infringement risk exists even if training is ultimately blessed as fair use: when the output itself is substantially similar to a specific copyrighted work. Generative models can absorb and reproduce stylistic choices across an artist’s whole body of work, and courts have historically struggled with the line between unprotectable style and protectable expression.
Penalties for AI-Related Infringement
If a court finds infringement, the numbers are serious. Copyright holders whose works were registered before the infringement can elect statutory damages instead of proving actual losses. The standard range is $750 to $30,000 per work, and a judge can raise a willful infringement award to as much as $150,000 per work.8Office of the Law Revision Counsel. 17 USC 504 – Remedies for Infringement: Damages and Profits
Large-scale commercial infringement can also draw federal criminal charges. Reproducing or distributing at least ten copies of copyrighted works with a total retail value over $2,500 within a 180-day period carries up to five years in prison.9Office of the Law Revision Counsel. 18 U.S. Code 2319 – Criminal Infringement of a Copyright Felony fines for an individual defendant can reach $250,000.10Office of the Law Revision Counsel. 18 USC 3571 – Sentence of Fine These penalties come from existing copyright law and apply to AI-related infringement the same way they apply to any other kind.
Protecting Your Own Work From Being Scraped
The legal tools available today for creators worried about AI training are limited. The robots.txt standard and “noai” metadata tags are meant to signal that a site’s content should not be scraped for training, but compliance is voluntary and there is no enforcement mechanism against a company that ignores them.
Researchers at the University of Chicago developed two technical tools that take a different approach. Glaze adds imperceptible changes to images that disrupt an AI model’s ability to mimic an artist’s style. Nightshade embeds distortions designed to corrupt a model’s training if the image is ingested. Both leave images looking normal to human viewers.
The most reliable legal protection is still registration. You cannot recover statutory damages or attorney’s fees unless your work was registered before the infringement, or within three months of publication. Proactive registration puts you in the strongest position if you later have to bring a claim.
New Federal Laws on AI-Generated Content
Congress has started legislating around AI content, though a comprehensive copyright statute for AI has not arrived. The most consequential new law is the TAKE IT DOWN Act, signed on May 19, 2025. It makes publishing non-consensual intimate images, whether authentic or computer-generated, a federal crime when done with intent to harm or without the subject’s consent. Platforms that host user content must remove such images within 48 hours of notice.11Congress.gov. S.146 – TAKE IT DOWN Act Penalties include prison time, fines, and mandatory restitution.
Broader federal protection for digital replicas of a person’s voice, image, and likeness is not yet law. The NO FAKES Act, introduced in April 2025, would create a federal right of publicity covering unauthorized AI-generated replicas in commercial contexts. As of mid-2026 that bill remains pending in the Senate Judiciary Committee.12Congress.gov. S.1367 – NO FAKES Act of 2025 Until it or something similar passes, protection against AI voice cloning and deepfakes outside intimate imagery relies on a patchwork of state right-of-publicity laws and, in some cases, federal unfair-competition claims under trademark law.
Commercial Use and Indemnification
If you plan to use AI-generated content commercially, the terms of service for your AI tool matter. Some major providers now offer contractual indemnification to enterprise customers, agreeing to cover legal costs and damages if an output is later found to infringe. The protections vary widely: some apply only to specific features, cap total liability at a negotiated amount, or require the customer to follow usage guidelines to stay eligible. Free-tier and individual-plan users rarely get any indemnification.
Before relying on AI output in a commercial product, read the terms carefully. Check whether the provider claims ownership or license rights over the output, whether third-party infringement claims are covered, and what conditions void that coverage. Providers update their terms often enough that a version from six months ago may not reflect where the company stands today.