AI copyright: how to protect AI-created content and what the law says

01The current state of AI copyright law

A copyright is a type of intellectual property that gives its owner the exclusive legal right to copy, distribute, adapt, display, and perform a creative work, usually for a limited time. The creative work may be in a literary, artistic, educational, or musical form. Copyright is intended to protect the original expression of an idea in the form of

Artificial intelligence has disrupted the foundations of copyright law. Copyright was designed to protect works of human authorship, and its core requirement, that a work be original and created by a human, is now being tested by generative AI tools that produce text, images, music, and code with varying degrees of human direction. The U.S. Copyright Office, the EU Intellectual Property Office, and courts in multiple jurisdictions are grappling with where to draw the line.

The current consensus in the United States is that purely AI-generated works are not copyrightable. The Copyright Office has stated repeatedly that works must contain a sufficient amount of human authorship to qualify for protection. However, the threshold for what constitutes sufficient human involvement remains unclear and is being litigated case by case.

02What can and cannot be copyrighted

Intellectual property (IP) is a category of property that includes intangible creations of the human intellect. There are many types of intellectual property, and some countries recognize more than others. The best-known types are patents, copyrights, trademarks, and trade secrets. The modern concep

Works entirely generated by AI without meaningful human creative input cannot be copyrighted. This means that an image produced by typing a short text prompt into a generative model, with no further editing or creative control, is generally not protectable. Anyone could copy, distribute, or adapt such a work without permission.

Works where a human exercises creative control over the output can be protected, but only the human-authored elements. If a writer uses AI to generate a first draft and then substantially edits, restructures, and rewrites the content, the final work may be copyrightable. The copyright would extend to the human creative contributions, not the AI-generated portions. The distinction is factual and must be assessed on a case-by-case basis.

03The human authorship requirement

Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and decision-making. It is a field of research in engineering, mathematics, and computer science that devel

The human authorship requirement has deep roots in copyright law. The U.S. Copyright Act of 1976 grants protection to original works of authorship, and the Supreme Court has interpreted authorship as a human creative act. The 2024 case of Thaler v. Perlmutter confirmed this principle when a federal court ruled that a work generated entirely by an AI system, with no human involvement, could not be registered.

The Copyright Office has applied this requirement in several high-profile registration decisions. In the case of the graphic novel Zarya of the Dawn, the Office registered the work but cancelled the registration for images that were generated by the Midjourney AI, preserving copyright only for the human-authored text and arrangement. This decision illustrates the granular approach the Office is taking, evaluating each element of a work separately.

04How to register AI-assisted works

The Copyright Office requires applicants to disclose AI-generated content in a registration. If a work contains material generated by AI, the applicant must identify those portions and clarify which parts were created by a human. Failing to disclose AI-generated content can result in cancellation of the registration, which means losing the ability to enforce copyright in infringement litigation.

The registration process involves filing an application with the Copyright Office, either online or by paper, depositing a copy of the work, and paying a fee. For AI-assisted works, the applicant should describe the human creative contributions in detail, such as selecting, arranging, editing, or modifying the AI output. The more specific the description of human involvement, the stronger the registration.

Creators should document their creative process. Keeping records of prompts, drafts, edits, and revisions provides evidence of human authorship if the registration is challenged. This documentation is especially important for works where the boundary between AI generation and human creation is blurred, as it demonstrates the extent of human creative control.

05The difference between AI-generated and AI-assisted

The distinction between AI-generated and AI-assisted works is the central legal question. An AI-generated work is one where the AI produces the creative output with minimal human direction, such as a single text prompt. An AI-assisted work is one where the human uses AI as a tool within a broader creative process, exercising meaningful creative judgment over the selection, arrangement, and modification of the output.

The Copyright Office guidance suggests that the more creative control a human exercises, the more likely the work is protectable. A photographer who uses AI to remove background clutter from an image they composed and shot is using AI as a tool, and the photograph remains their original work. A person who generates an image entirely from a text prompt and publishes it unchanged is likely creating an unprotectable AI-generated work.

This spectrum creates practical uncertainty. A musician who co-writes lyrics with an AI assistant, a programmer who uses a code generator and then refactors the output, and a graphic designer who uses AI to generate options and then manually composites them into a final design are all on different points of the spectrum. The legal treatment of each depends on the specific facts.

06What the courts have ruled

Several key cases have shaped the AI copyright landscape. In Thaler v. Perlmutter, a federal court ruled that works generated by AI without human authorship cannot be copyrighted. In Andersen v. Stability AI, artists sued AI image generators for training on copyrighted works without permission, raising questions about whether training constitutes infringement. The case is ongoing and could establish precedent for how AI training data is treated.

The New York Times v. OpenAI lawsuit raises similar questions about text training data. These cases may determine whether using copyrighted works to train AI models constitutes fair use or requires licensing. The outcomes could reshape the economics of AI development, as licensing requirements would increase costs for model training.

Beyond training data, courts are also addressing infringement by AI output. If an AI model generates an image that closely resembles a copyrighted work, is that infringement? The answer depends on whether the output is substantially similar to the protected work and whether the AI had access to it during training. These questions are being litigated and will likely produce a body of case law over the coming years.

07How creators should approach AI copyright

Creators using AI tools should take several practical steps. First, disclose AI-generated content in any copyright registration to avoid losing the registration. Second, document the creative process with records of prompts, edits, and human creative decisions. Third, exercise meaningful creative control over the final work rather than publishing raw AI output.

For commercial works, creators should consider whether they need protection at all. If a work is entirely AI-generated and therefore uncopyrightable, competitors can legally copy it. This may be acceptable for some uses, but for works with commercial value, human creative contributions are necessary to obtain legal protection. The safest approach is to use AI as a tool within a process that includes substantial human creativity.

The legal landscape is evolving rapidly. Creators should monitor Copyright Office guidance, court decisions, and legislative developments. The intersection of AI and copyright is one of the most actively litigated areas of intellectual property law, and rules that apply today may change as new cases are decided and new regulations are issued.