Who Owns the Rights to AI-Generated Images? A Straightforward Guide to Copyright Law

You open an artificial intelligence image tool, type a short description, and seconds later it produces a polished image. It might look like a magazine cover or a marketing graphic.

So, who owns it?

The answer under current U.S. law is simple. If the image is generated by AI without meaningful human creative control, no one owns the copyright. The image falls into the public domain. Anyone can use it, copy it, or sell it.

That result surprises people, but it follows a long-standing rule.

Human Authorship Is Still Required

U.S. copyright law protects “original works of authorship,” and courts have consistently read that phrase to require a human creator. The Copyright Office has applied that rule for years, and it has not changed with the rise of artificial intelligence.

That principle was tested by Stephen Thaler, who sought to register an image created entirely by his AI system. The courts rejected the claim, and in March 2026 the Supreme Court of the United States declined to review the case. The lower court rulings remain in place, and the human-authorship requirement stands.

The takeaway is simple. Copyright protects human authorship, and it does not extend to works created entirely by a machine.

Why Prompts Alone Are Not Enough

Many AI users assume that writing a detailed prompt makes them the author. The law does not see it that way.

The Copyright Office’s 2025 guidance explains that a prompt is closer to giving instructions than creating the work itself. The AI system determines the composition, style, and details. That means the AI, not the user, performs the core creative work.

As a result, an image generated from a prompt alone does not qualify for copyright protection.

What Counts as Enough Human Input?

You can still obtain copyright protection, but only for the parts of the work that reflect your own creative decisions.

The key question is whether you exercised real creative control over the final image.

That usually requires more than generating a single output. Examples that tend to qualify include:

• Substantial editing after generation, such as detailed work in Photoshop
• Selecting, arranging, and combining multiple AI images into a new composition
• Iterative prompting paired with meaningful manual changes to the result
• Blending AI elements with your own photographs, drawings, or design work

The Copyright Office’s decision in Zarya of the Dawn illustrates the point. That case involved a comic book where the author wrote the story but used an AI tool to generate the images. The creator received protection for the text and the arrangement of the book, but the individual AI-generated images were not protected.

That approach reflects how the Copyright Office treats these works in practice. It will register only the portions of a work that reflect human creativity and will exclude material generated by AI. If you apply for registration, you must disclose any AI use and claim only the human-created portions.

Platform Terms Are Not Copyright

Many platforms try to simplify this issue through their terms of service. For example, OpenAI’s terms assign users the company’s rights in the images they generate, and Adobe takes a similar approach with its Firefly system, particularly for commercial use.

Those rights, however, come from a contract, not from copyright law. That distinction matters. A contract governs how you and the platform can use the image, but it does not create the exclusive rights that copyright provides.

In practical terms, that means you may have permission to use and even sell a purely AI-generated image under the platform’s rules. But if the image lacks human authorship, you do not have the right to stop others from using the same or a similar image.

Practical Steps for Businesses and Creators

If you plan to use AI-generated images in a commercial setting, a few basic habits will put you in a stronger position. Start by documenting your process. Save your prompts, drafts, and edits so you can show where your own creative input shaped the final image. That record can matter if you later seek registration or need to defend your rights.

It also helps to treat AI as a tool rather than the finished product. Add your own material and make real design choices so the end result reflects your judgment, not just the system’s output. At the same time, read the platform’s terms before you rely on the image. Commercial rights vary, and those terms often control how you can use what you generate.

You should also be prepared to disclose AI use where required. As noted above, the U.S. Copyright Office requires disclosure in registration applications, and many marketplaces and stock platforms now ask for it as part of their listing process. And keep an eye on infringement risk. Ongoing disputes involving The New York Times Company and Getty Images show that questions about training data and output similarity are still being worked out in the courts.

For projects with real commercial value, it is worth getting advice before you publish or try to register the work. A short review on the front end is usually cheaper than sorting out a dispute later.

Bottom Line

Pure AI-generated images do not receive copyright protection in the United States. That rule has held steady, even as the technology has improved.

You gain real protection only when your own creative contribution shapes the final work.

The practical approach is simple. Use AI to speed up the process, but make sure the final product reflects your own judgment and effort. That is what the law protects.

Jeff Wilson is a Pender & Coward attorney focusing his practice on employment law, litigation, local government, and intellectual property matters.