What to Expect When You Want to Apply for a Trademark

“ I made my logo with AI in about thirty seconds. Can I actually protect it? ” I hear some version of this more and more. The short answer is probably yes. The reason why says a lot about the difference between trademark and copyright. Trademark Doesn’t Care Who (or What) Made It In our post on AI and copyright , we covered Thaler v. Perlmutter, the case confirming that copyright requires human authorship. Content generated without meaningful human input generally can’t be copyrighted. Trademark asks a different question. It isn’t about how a mark was created. It’s about what the mark does in the marketplace: whether customers see it and associate it with your business. The USPTO doesn’t ask whether a person, a designer, or an AI tool came up with your AI-generated brand name or logo. It asks three things: Does it identify your business as the source of your goods or services? Is it distinctive, rather than generic or merely descriptive? Does it avoid a likelihood of confusion with someone else’s existing mark? If the answers are yes, an AI logo trademark application stands on the same footing as any other. Where AI Logos Run Into Trouble The problem usually isn’t the AI. It’s what AI tools tend to produce. They draw on enormous datasets, so they often generate designs that resemble existing marks or lean on generic industry imagery. Type “cloud” or “data” into a logo generator and you’ll see the same icons and circuit patterns again and again. The tool isn’t checking the USPTO database before it hands you a design. That makes a clearance search more important when a logo came from AI, not less. A “unique” logo that turns out to be confusingly similar to a registered mark can mean a refused application, or worse, a cease-and-desist letter after you’ve printed the signage. What You’ll Need to Show A few practical points if you want to trademark an AI-generated logo: Real use in commerce. The USPTO expects to see the mark actually in use, such as on packaging, a live website, or advertising. AI mockups or “coming soon” pages are not sufficient. A person or legal entity as owner. Only people and registered businesses can own a trademark, never the AI tool. Distinctiveness. A generic name paired with a generic icon gives you very little legal strength, however polished it looks. The Copyright Wrinkle Here’s where the two areas of law collide. Even after your AI-generated logo is trademarked, the artwork itself may not be copyrightable, per the same human-authorship rule from Thaler. Trademark stops others from using your mark to identify competing goods or services, but it doesn’t automatically stop someone from copying the design in every context. If a human designer meaningfully edits or builds on the AI output, that human contribution may carry its own copyright protection on top of the trademark. That’s a good reason to have a person put real creative work into the final version rather than using the raw AI output as-is. Before You Build a Brand Around It If you’ve generated a logo or name with AI and you’re about to spend real money on signage, packaging, and marketing, the order of operations matters: clearance search first, document your actual use, and then file. Skipping the search is the most common way an AI-generated brand identity turns into an expensive rebrand a year later. Whether you’re protecting a new logo or sorting out trademark AI-generated content more broadly, brand protection starts before you launch, not after someone else files first. Contact Rosenthal IP Law today to schedule a consultation on trademarking your AI-generated logo or brand name before the signage goes up.

If your team has been leaning on AI tools to generate blog posts, product images, or social content, here's a question worth asking before you publish the next one: who actually owns that content? For a lot of businesses, the honest answer is nobody, and that's a bigger problem than it sounds. The Basic Question The U.S. Copyright Office has been studying this since 2023, collecting public input and issuing guidance in stages. Its conclusion has stayed consistent throughout: copyright protection requires human authorship, and that requirement hasn't changed just because the tool doing the work became a lot more sophisticated. The case Thaler v. Perlmutter addressed this at the Federal level. A computer scientist generated an artwork entirely through an AI system and applied to register a copyright for the artwork, listing the AI system itself as the author. The Copyright Office refused registration, the D.C. Circuit Court of Appeals upheld that refusal in 2025, and on March 2, 2026, the Supreme Court declined to hear the case. That left the ruling standing: human authorship is a "bedrock requirement" of copyright, and content generated without meaningful human involvement simply falls outside of copyrightability. The harder, more practical question, one the Copyright Office has addressed in detail, is what happens when a human is the one typing the prompts. Its answer: prompts alone, no matter how detailed, generally don't make the person who wrote them the "author" of the output. A prompt functions more like an instruction than an expression, since the AI system is still the one making the real expressive choices, what the final image looks like, how a sentence is phrased, in between the prompt and the result.