Can You Copyright AI-Generated Content?
The short answer: it depends on how much human creativity went into the work.
Generative AI tools can now produce text, images, code, music, and video that is often indistinguishable from human-created work. For Illinois businesses using these tools to generate marketing copy, software code, design assets, or technical documentation, the question of who owns the output is not academic — it affects whether you can stop a competitor from copying your work.
The good news is that we now have a clear framework. In January 2025, the U.S. Copyright Office published Copyright and Artificial Intelligence, Part 2: Copyrightability, a comprehensive report analyzing exactly what kind of human involvement is needed to bring AI-generated outputs within the scope of copyright protection. In March 2025, the D.C. Circuit affirmed the human authorship requirement in Thaler v. Perlmutter. And in March 2026, the U.S. Supreme Court declined to hear the case, leaving that ruling intact.
Together, these developments give businesses a workable set of rules. Here are the key points.
Key Takeaways
Purely AI-generated content is not copyrightable. If an AI system autonomously creates a work without meaningful human creative input, nobody owns a copyright in it. The Supreme Court's refusal to review the Thaler decision in March 2026 effectively settles this question for now.
Using AI as a tool does not cost you your copyright. If a human author uses AI to assist with grammar checking, formatting, brainstorming, code debugging, or similar tasks, the resulting human-authored work is fully protectable — just as it would be without the AI.
Prompts alone are not enough. The Copyright Office concluded that text prompts — no matter how detailed or how many times you revise them — do not give a user enough control over the AI's expressive output to qualify as authorship.
Your own creative work fed into an AI system can be protected. When a human inputs their own copyrightable work (a drawing, a photograph, written text) and that work remains visible in the AI's output, the human's original expression is still protected.
Human editing and arrangement of AI output can create copyright. If you substantially modify, edit, or creatively arrange AI-generated material, your contributions can be copyrightable — even though the raw AI output underneath is not.
Contracts fill the gaps. Where copyright protection is uncertain, well-drafted agreements specifying ownership, confidentiality, and acceptable AI use become your primary defense.
The Human Authorship Requirement
U.S. copyright law protects "original works of authorship." The Copyright Office and federal courts have consistently interpreted "authorship" to require human creative involvement. This principle has deep roots — the Copyright Clause of the Constitution grants Congress the power to secure rights to "Authors," and the Supreme Court has interpreted this to mean human beings. As the Supreme Court explained in Community for Creative Non-Violence v. Reid, "the author [of a copyrighted work] is . . . the person who translates an idea into a fixed, tangible expression entitled to copyright protection."
No court has recognized copyright in material created by non-humans. The Ninth Circuit has rejected copyright claims for text purportedly "authored by non-human spiritual beings" and for photographs taken by a monkey. These holdings confirm that copyright's reach begins and ends with human creativity.
Thaler v. Perlmutter: The Case That Went to the Supreme Court
The foundational case applying this principle to AI is Thaler v. Perlmutter. Computer scientist Stephen Thaler developed an AI program called the "Creativity Machine" and used it to produce a visual artwork titled "A Recent Entrance to Paradise." Thaler listed the AI system as the sole author and claimed ownership through a work-for-hire theory. He did not prompt the AI system himself or make any edits to the final image.
The Copyright Office refused to register the work. In August 2023, the U.S. District Court for the District of Columbia upheld that refusal, holding that "copyright law protects only works of human creation" and that "human authorship is a bedrock requirement of copyright."
In March 2025, the D.C. Circuit affirmed, holding that human authorship is required "as a matter of statutory law" under the Copyright Act. The court was careful to note that its ruling does not prohibit copyrighting work made by or with the assistance of artificial intelligence — only that the AI itself cannot be the author.
In March 2026, the U.S. Supreme Court denied certiorari, declining to review the case. That denial leaves the D.C. Circuit's ruling intact and effectively settles the question: an AI system cannot be an "author" under federal copyright law.
The Copyright Office's January 2025 Report
Even before the Supreme Court acted, the Copyright Office had already provided detailed guidance. Its January 2025 Part 2 Report drew on more than 10,000 public comments and reached several conclusions that matter for businesses.
The report confirmed that existing law is adequate to resolve AI copyrightability questions — no new legislation is needed. It reaffirmed that using AI as an assistive tool does not undermine copyright protection. And it set out a framework for analyzing the three types of human contribution that can make an AI-assisted work copyrightable: prompts, expressive inputs, and modifications or arrangements of AI-generated content.
The Three Types of Human Contribution
The Copyright Office's framework organizes human involvement with AI into three categories. Understanding them is the most practical thing a business owner can do to protect AI-assisted work product.
1. Prompts — And Why They Are Not Enough
A prompt is an instruction — usually text — that tells an AI system what to generate. Prompts can range from a few words ("cartoon spaceship") to detailed paragraphs specifying subject matter, style, composition, and technique.
The Copyright Office concluded that, based on current technology, prompts alone do not provide sufficient human control to make the user an author of the output. The reasoning is straightforward: prompts function as instructions that convey unprotectable ideas, but they do not control how the AI system processes those instructions into expression. The gap between what you ask for and what the AI produces is too wide — the system fills in the expressive details on its own.
The Office illustrated this with a concrete example. It entered a detailed prompt into a commercially available AI system describing a "bespectacled cat in a robe reading the Sunday newspaper and smoking a pipe" with specifications for lighting, composition, and style. The resulting image reflected some of those instructions, but the system independently decided the cat's breed, coloring, pose, facial expression, and clothing. The same prompt, run again, produced a substantially different image. That unpredictability demonstrates a lack of human control over the expressive elements.
What about iterative prompting? The Copyright Office specifically addressed this and concluded that submitting revised prompts multiple times is essentially "re-rolling the dice" — causing the system to generate more outputs from which to choose, but not increasing the user's control over the creative expression in any particular output. Copyright protects original authorship, not effort or persistence.
What about selecting among multiple outputs? Also not enough. The Office agreed that choosing a favorite from among uncontrolled AI-generated options is not itself a creative act sufficient for copyright. As one commenter put it, walking into a gallery and finding a painting that matches an idea you already had does not make you the author of that painting.
2. Expressive Inputs — The Strongest Path to Protection
This category is where many businesses will find the strongest path to copyright protection. Some AI systems allow users to input their own creative work — a drawing, a photograph, a musical composition, a block of written text — and instruct the system to modify, enhance, or build on it.
When a human inputs their own copyrightable work into an AI system and that work remains perceptible in the output, the human is the author of at least that portion of the result. Their original creative expression is protected, with a scope similar to protection in a derivative work. The copyright covers the perceptible human expression and may also cover the selection, coordination, and arrangement of human-authored and AI-generated material — though it does not extend to AI-generated elements standing alone.
The Copyright Office highlighted a registered work called Rose Enigma to illustrate this point. The artist created a hand-drawn illustration of a masked face with rosebuds and used it as an input to an AI system along with a text prompt. The output clearly retained the expressive elements of the original drawing — the outline of the mask, the position of the nose, mouth, and cheekbones, the arrangement of stems and rosebuds, and the shape and placement of leaves. The Office registered the work, limiting the registration to "unaltered human pictorial authorship that is clearly perceptible in the deposit and separable from the non-human expression that is excluded from the claim." The AI-generated enhancements — the realistic three-dimensional rendering, lighting, and shadows — were not covered.
By contrast, the Office upheld a refusal to register a photograph that had been processed through an AI app trained on Van Gogh's The Starry Night. Unlike Rose Enigma, the copyrightable input was not clearly perceptible in the output, and the applicant had exerted insufficient creative control over the AI's generation process. The lesson: inputting your own work into an AI system only helps if your original expression survives in a recognizable way.
3. Modifying or Arranging AI-Generated Content
Generating content with AI is often a first step, not a final product. The Copyright Office confirmed that human authorship added after AI generation can support copyright protection.
A human who selects, coordinates, and arranges AI-generated material in a sufficiently creative way can claim copyright in the resulting work as a whole — even though the individual AI-generated components are not separately protectable. This is the same principle that protects compilations: the creativity lies in the selection and arrangement, not in the underlying pieces.
The Zarya of the Dawn decision is the leading example. The Office found that the author's choices about which AI-generated images to use, where to place them, and how to combine them with her own text constituted protectable authorship in the work as a compilation.
A human can also modify AI-generated material — editing, adapting, enhancing, or rewriting it — to a degree that the modifications themselves meet the originality standard. The copyright in that case would cover the human's contributions but not the underlying AI-generated content.
Many popular AI platforms now offer tools that encourage this kind of iterative human involvement. Features like region selection, inpainting, and remix prompting allow users to select specific areas of an AI-generated image and regenerate them with different instructions. Whether these modifications meet the originality standard will depend on the specific facts of each case.
Importantly, the inclusion of AI-generated elements within a larger human-authored work does not undermine the copyrightability of the larger work. A film that includes AI-generated special effects or background artwork remains copyrightable, even if the AI-generated elements are not separately protectable.
Is AI-Assisted Work Still Copyrightable?
Yes — and this is worth emphasizing, because it is the source of most confusion. The Copyright Office drew an important line that should reassure businesses already using AI day to day: there is a difference between using AI as a tool to assist in creating works and using AI as a stand-in for human creativity.
Assistive uses — grammar checking, spell checking, formatting, code debugging, copy editing, brainstorming, generating outlines for reference — do not limit copyright protection for the resulting work. If a human author writes an article and uses AI to check grammar or suggest word choices, the article remains the human's copyrightable work. If a musician uses AI to identify chord progressions or a filmmaker uses AI for color correction, the human-authored work is unaffected.
Using AI as a brainstorming tool — for example, prompting an AI system to help generate ideas or create a preliminary outline, then writing the work yourself without incorporating the AI's actual output — also does not affect copyrightability. The key is that the human is referencing the AI's output for inspiration, not incorporating it as part of the final work.
The distinction matters when AI crosses the line from assistance to creation — when the AI system is making the expressive choices rather than the human.
What About AI-Generated Code?
AI code generation tools — GitHub Copilot, Claude, ChatGPT, and their successors — are now standard parts of many development workflows. The Copyright Office's framework applies to code just as it applies to images, text, and music.
When a developer designs the architecture, specifies the logic, writes the core implementation, and uses AI to handle mechanical tasks like boilerplate generation, auto-completion, or syntax correction, the developer's work is likely protectable. The AI is functioning as an assistive tool, and the developer's creative choices drive the final product.
When a developer enters a high-level description and accepts the AI's complete output with minimal review or modification, the analysis becomes more difficult. Under the Copyright Office's framework, the developer's contribution looks more like a prompt — conveying an idea but not controlling the expressive details of the implementation. The resulting code may not be protectable, or may be protectable only to the extent the developer substantially modified or arranged the AI-generated output.
Open Source License Risks
AI code generation tools are trained on large repositories of code, including open-source code governed by various license terms (GPL, MIT, Apache, and others). There is a recognized risk that AI tools may reproduce or closely paraphrase code from their training data. If an AI tool generates output that substantially reproduces GPL-licensed code, for example, the resulting code could be subject to GPL obligations — potentially requiring disclosure of proprietary source code.
Companies should implement review processes for AI-generated code, particularly for code that will be distributed or included in commercial products. Automated tools that scan for code similarity can help identify potential license conflicts.
Trade Secret Considerations
Inputting proprietary code or confidential business information into third-party AI tools raises trade secret concerns. If confidential information is used to train or improve a third-party model, you may have compromised your trade secret protections. Review your AI vendor agreements carefully, and consider using enterprise-tier AI tools that contractually commit to not training on your data.
How Are Other Countries Handling AI Copyright?
The Copyright Office's January 2025 report surveyed international approaches, and for businesses operating across borders, the global landscape matters.
South Korea issued guidance in 2023 confirming that only a natural person can become an author and that copyright registration for an AI output is not available unless a human contributed creatively to the expressive form. Creative selection and rearrangement of AI output can support registration as a compilation.
Japan published guidelines in May 2024 stating that copyrightability of AI-generated content will be determined case by case, considering the amount and content of the user's prompts, the number of generation attempts, whether the user selected from multiple outputs, and any subsequent human additions and corrections.
China saw its Beijing Internet Court find in 2023 that an image created using Stable Diffusion was protected under Chinese copyright law, emphasizing that the user had selected over 150 prompts and made subsequent adjustments. This analysis reached a different conclusion than the U.S. Copyright Office regarding the sufficiency of prompting, reflecting a genuine divergence in approaches.
The European Union member states largely agreed in a 2024 policy questionnaire that current copyright principles are adequate, that purely AI-generated works cannot be protected, and that AI-generated content may qualify for copyright only if the human input was significant.
The United Kingdom has a unique statute, predating generative AI, that provides limited protection for "computer-generated works" where there is no human author. This approach is an outlier among major jurisdictions.
The common thread: purely AI-generated content, without meaningful human creative involvement, is not protectable in virtually any major jurisdiction.
Illinois-Specific Considerations
Illinois businesses should be aware that the state has moved ahead of most jurisdictions on related AI issues, even if copyright law itself is federal.
In August 2024, Governor Pritzker signed HB 4875 into law, amending the Illinois Right of Publicity Act to specifically address generative AI. Effective January 1, 2025, the law makes it illegal to produce or distribute unauthorized digital replicas of a person's voice, image, or likeness using AI. Illinois became the second state in the nation — after Tennessee — to enact these protections.
Separately, HB 4762, the Digital Voice and Likeness Protection Act, restricts the use of digital replicas in employment contracts. A contract provision allowing an employer to create a digital replica of a worker's voice or likeness is unenforceable unless the contract specifically describes the intended uses and the worker was represented by counsel or a union.
These laws do not directly address copyright in AI-generated content — that remains governed by federal law. But they create additional obligations and potential liability for Illinois businesses using AI to generate content that involves a real person's identity. If your marketing team uses AI to create images or audio that resemble specific individuals, these state laws apply on top of the federal copyright analysis discussed above.
Practical Steps for Illinois Businesses
Document Human Involvement
If you use AI tools in your creative or development workflow, document the human creative contributions at each stage. Keep records of the human-authored inputs provided to the AI (drawings, photographs, text, code), the modifications, selections, and editorial decisions made by humans after AI generation, and the iterative process of refinement and revision.
This documentation supports copyright registration and demonstrates the human authorship that forms the basis of any copyright claim.
Structure Workflows for Maximum Protection
Based on the Copyright Office's framework, the strongest copyright position comes from workflows where humans provide their own expressive inputs to AI systems and retain perceptible creative expression in the output, or where humans substantially modify and arrange AI-generated material rather than using it as-is. Simply entering prompts and selecting outputs is the weakest position. Design your creative and development processes with this hierarchy in mind.
Use Contracts to Fill the Gaps
Where copyright protection is uncertain, contractual protections become essential. Your agreements with clients, vendors, employees, and contractors should specify ownership of AI-assisted work product, address confidentiality obligations for information shared with AI tools, allocate risk for potential IP issues in AI-generated outputs, and define acceptable use of AI tools in the performance of work.
Implement an AI Usage Policy
Companies should establish clear internal policies governing which AI tools are approved for use, what types of information can be input into AI tools, review and approval processes for AI-generated work product, documentation requirements for human contributions, and compliance with Illinois's digital replica laws when AI-generated content involves real individuals.
Register Strategically
The Copyright Office allows registration of works that contain AI-generated elements, provided the applicant discloses those elements and demonstrates sufficient human authorship in the work as a whole. Transparency about the AI's role is essential — the Office has canceled registrations where applicants failed to disclose AI involvement. If your work contains more than a de minimis amount of AI-generated material, identify it in your application and describe your human contributions clearly.
Frequently Asked Questions
Is AI-generated content in the public domain? Purely AI-generated content — created without meaningful human creative involvement — is not eligible for copyright protection under current U.S. law. That means anyone can copy, use, or distribute it without permission. However, if a human substantially modifies or creatively arranges AI-generated material, the human's contributions can be copyrighted.
Can I copyright something I made with ChatGPT or Midjourney? It depends on your level of creative involvement. If you entered a prompt and used the output without significant changes, that output likely is not copyrightable. If you used the AI-generated content as raw material and substantially edited, arranged, or combined it with your own creative work, your contributions may be protectable.
Does using AI tools mean I lose copyright in my work? No. The Copyright Office has been clear that using AI as an assistive tool — for grammar checking, brainstorming, formatting, code completion, or similar tasks — does not affect your copyright in the resulting work. The concern arises only when AI is generating the expressive content rather than assisting a human who is.
Do I have to tell the Copyright Office if I used AI? Yes, if your work contains more than a de minimis amount of AI-generated material. The Copyright Office requires applicants to disclose AI-generated content and describe the human author's contributions. Failure to disclose can result in cancellation of the registration.
What about AI-generated content that uses someone's likeness? In Illinois, this triggers additional state-law obligations. HB 4875, effective January 1, 2025, prohibits the unauthorized production or distribution of digital replicas of a person's voice, image, or likeness created using AI. Violations can result in liability separate from any copyright issues.
How does this affect my business's AI-generated marketing materials? If your marketing team uses AI to generate copy, images, or video, treat the AI output as a starting point — not a finished product. Have humans substantially edit and refine the material, document those contributions, and ensure that AI-generated content involving real people complies with Illinois's digital replica laws.
Looking Ahead
Copyright law in the AI era continues to evolve. The Copyright Office has signaled that it will issue additional registration guidance and update its Compendium of U.S. Copyright Office Practices. A subsequent part of the Office's AI report will address the training of AI models on copyrighted works, licensing considerations, and allocation of liability — issues that may further reshape the landscape.
But the fundamental principle is now settled at every level of the federal courts: human creativity is the foundation of copyright protection. Businesses that build their workflows around this principle — maximizing human creative involvement, providing their own expressive inputs, documenting their contributions, and using contracts to address gaps — will be best positioned regardless of how the law develops.
Need Legal Help?
If your business uses AI tools to generate content, code, or creative works, The Law Office of Krista Krepp can help you develop policies and contractual frameworks that protect your interests. We offer free consultations for technology and IP matters. Contact us at contact@krepplaw.com or call 312-292-9887.