If you’ve used ChatGPT, Midjourney, or Claude to create anything—articles, images, code, marketing copy—you’ve probably wondered: do I own this? Can I protect it? The short answer is complicated, and it matters more than you think.
Copyright law wasn’t written for a world where machines generate creative work. Courts, copyright offices, and legislators are still figuring it out. But if you’re using AI tools professionally, you need to understand where the law stands right now and what that means for your work.
The U.S. Copyright Office’s Current Stance
The U.S. Copyright Office has been consistent: work created entirely by AI, without human authorship, cannot be copyrighted. This isn’t a new policy—it’s rooted in the requirement that copyright protects works of human authorship.
In practice, this means if you type a prompt into Midjourney and use the output as-is, that image isn’t copyrightable. Same goes for text generated by ChatGPT or Claude with minimal human input. The Copyright Office has rejected multiple applications for purely AI-generated works, and courts have upheld this position.
But here’s where it gets interesting: the Office has also said that works involving AI can be copyrighted if there’s sufficient human creativity involved. The question is what counts as “sufficient.”
What Counts as Human Authorship?
The line between “AI-generated” and “human-authored with AI assistance” is blurry, but emerging case law and guidance offer some clarity.
Generally protected:
- Significant editing, arrangement, or modification of AI output
- Using AI as one tool in a larger creative process (e.g., AI drafts that you substantially revise)
- Selecting and combining AI-generated elements in creative ways
- Original prompts combined with heavy post-processing (especially for images)
Generally not protected:
- Raw AI outputs with no human modification
- Minor tweaks or corrections to AI-generated content
- Simply choosing between AI-generated options
- Prompt engineering alone, no matter how sophisticated
A useful analogy: a photographer owns the copyright to their photos, even though the camera does the technical work. But if you set up a camera on a timer and it randomly captures an image, that’s different. The question is always whether you made creative choices that shaped the final work.
What This Means for Different Use Cases
Writers and content creators: If you’re using AI to draft and then heavily editing—restructuring, rewriting, adding analysis—you likely have a strong copyright claim. If you’re publishing AI outputs with light touch-ups, you’re in murkier territory.
Designers and artists: Midjourney or DALL·E images used as-is are risky. But if you’re using AI to generate elements that you then composite, modify, or incorporate into larger works, you’re building a case for human authorship.
Developers: Code is particularly interesting. Copyright protects the expression of code, not functionality. AI-generated boilerplate probably isn’t copyrightable, but substantial original architecture and implementation—even if assisted by GitHub Copilot—likely is.
Businesses: If you’re selling or licensing content, the lack of clear copyright protection is a business risk. Competitors could legally copy purely AI-generated work. Many companies are now requiring employees to document their creative process when using AI tools.
The International Puzzle
U.S. law is just one piece. The EU, UK, Japan, and other jurisdictions are taking different approaches. Some countries offer more protection for AI-generated works; others are stricter. If you’re working internationally or for global clients, the inconsistency creates additional complexity.
China, for example, has ruled in some cases that AI-generated content can be copyrighted if someone made creative choices in the process. The UK has considered (but not yet enacted) special protections for computer-generated works.
Practical Steps to Protect Your Work
While the law evolves, here’s what you can do now:
- Document your process. Save drafts, notes, and revision history showing your creative input.
- Add substantial human creativity. Don’t publish raw AI outputs. Edit, revise, and add original elements.
- Understand your tools’ terms. Most AI platforms grant you rights to outputs, but some have restrictions on commercial use or attribution.
- Consider contractual protection. If copyright is uncertain, use contracts, NDAs, or trade secret protection for sensitive work.
- Stay informed. This area is changing fast. Court decisions in 2026 could reshape what’s possible.
The bottom line: AI is a powerful tool, but purely automated output won’t get you legal protection. The more human creativity you bring to the process, the stronger your claim. As these tools become more capable, expect the legal framework to keep evolving—probably for years to come.
Want to stay ahead of shifts like this? Subscribe to the One Two Three AI newsletter—one useful AI idea delivered daily, so you can keep up without the noise.