5 min read
When the U.S. Copyright Office asked the public what it thought about AI and copyright, it received more than 10,000 comments. That flood says something: almost everyone who makes things for a living has an opinion on who owns AI-made work, and a lot of those opinions are wrong.
For a solo owner, this is not an abstract debate. It decides whether you can stop a competitor from copying the logo you made in an image generator, whether the product photos on your shop are really yours, and what you should tell a client who asks if they own the design you delivered. Here are five common beliefs, tested against what the Copyright Office and the courts have actually said.
A quick note: this is general information about U.S. law, not legal advice. Rules differ by country, and a short call with an intellectual property lawyer is worth it before you build a brand on AI-made assets.
Myth 1: “If I wrote the prompt, I own the picture”
Verdict: mostly false.
In January 2025 the Copyright Office published Part 2 of its report on copyright and AI, which deals with exactly this question. Its conclusion is that prompts alone generally do not give you enough control over the output to make you its author. The reasoning is intuitive once you hear it: you can type the same prompt twice and get two very different images, so the specific picture came from the machine, not from you.
A long, clever prompt does not change that. Writing instructions is closer to commissioning work than to making it.
Myth 2: “Nobody can ever protect anything that touches AI”
Verdict: false, and this is the good news.
The same report is clear that using AI as a tool does not wipe out your rights. Human contributions can still be protected, including:
- Your own original material that shows up in the output, such as a sketch you drew and fed in.
- Creative selection and arrangement of AI pieces, like the way you combine images and text into a layout.
- Meaningful edits you make by hand after generation.
The well-known example is the comic book Zarya of the Dawn. The Copyright Office let the author keep protection for her writing and for how she arranged the pages, but not for the individual Midjourney images. In practice, the more of your own hand is in the final piece, the more of it you can protect.
Myth 3: “The courts will probably flip this soon”
Verdict: unlikely in the near term.
The biggest test case so far went the other way. In March 2025, the U.S. Court of Appeals for the D.C. Circuit ruled in Thaler v. Perlmutter that copyright requires a human author, and upheld the refusal to register an artwork listed as created by an AI system. That case involved a machine credited as the sole author, which is more extreme than most small business uses, but it shows where the starting point sits.
Policy can change, and the Copyright Office keeps its AI resource page updated as new parts of its study come out. But building your brand on the hope of a future ruling is a bet, not a plan.
Myth 4: “If I cannot copyright it, I cannot use it”
Verdict: false.
This one causes needless panic. Not being able to register copyright in an image does not mean you are forbidden from using it. You can generally still put an AI-made image on your website, in a social post, or on a flyer, subject to the terms of the tool you used. Most major image tools, including the ones built into Canva and ChatGPT, let paying users use outputs commercially, though you should read the current terms for your plan.
What you lose is exclusivity. If a competitor copies your AI-made hero image, you may have no copyright claim to stop them. That is fine for a blog header. It is a real problem for a logo or a product design that is supposed to be uniquely yours.
Myth 5: “Clients will never ask, so it does not matter”
Verdict: risky.
Freelancers increasingly see contracts that ask them to promise the work is original and that the client will own it outright. If a meaningful part of that work came straight out of an image generator, that promise may be hard to keep. The same goes the other way: if you hire a designer for your logo, you want to know how much of it a human actually made.
Detection is getting easier, too. We covered how new watermarking works in our myth-busting look at OpenAI’s text watermark. The safer path is to be upfront about what was AI-assisted and price accordingly.
A simple rulebook that holds up
None of this means you should stop using AI for visuals. It means matching the tool to how much ownership the asset needs:
- Disposable assets (social posts, blog headers, mood boards): use AI freely. Exclusivity barely matters.
- Brand assets (logo, mascot, signature product designs): use AI for ideas, then have a human redraw or substantially rework the final version, and keep the drafts that show your process.
- Client deliverables: add a line to your contract saying which tools you use and that purely AI-generated elements may not be protectable by copyright.
- Registration: if you do register a work with AI-made parts, the Copyright Office expects you to disclose them and claim only the human-made portion.
If you do not yet have written rules for how your business uses AI, our one-page AI policy guide has a section you can adapt for creative work. And if you make graphics in Canva, our Canva Craft review shows which kinds of images its new model handles best.
The skeptic’s bottom line
AI-made images are usable, but they are not automatically yours to defend. The more something matters to your brand, the more human work it needs before you call it your own. Have you ever had a client ask whether your work was AI-made? Tell us how you answered in the comments.
