Did you write a detailed prompt?
Generic prompts produce outputs no one owns. Specific, considered direction is the first evidence that the work reflects your judgment — not the model's.
Two separate questions, answered on two separate tracks: is it copyrightable, and is its name trademarkable? The law treats these differently, and so do we.
U.S. copyright requires human authorship. We measure how much of the character reflects your creative judgment, across five factors.
Trademark doesn't care who (or what) made the design. It protects names that identify a brand in commerce. Three factors decide whether yours qualifies.
U.S. copyright requires human authorship. Each factor measures a specific way your work might qualify — weighed together, not scored in isolation.
Generic prompts produce outputs no one owns. Specific, considered direction is the first evidence that the work reflects your judgment — not the model's.
Choosing between generations is a recognized pathway in USCO guidance, but a partial one. Selection alone rarely secures protection — it builds the case.
The strongest single factor. Painting, redrawing, compositing, or meaningfully editing the output is the clearest signal of authorship in the current USCO framework.
Text elements are independently copyrightable even when the image isn't. A strong name and written world can carry protection the visual alone can't.
The core test. If the output reflects a specific mental conception you directed into being, the work is authored — even if a model did the rendering.
Trademark doesn't require human authorship. It protects names used in commerce. If you're selling merch, this is the layer that keeps someone else from shipping yours.
Descriptive names ("Fast Delivery") don’t register. Coined or unexpected names ("Kodak," "Apple" for computers) do. This is the single biggest predictor of registrability.
Trademark rights flow from use in commerce. Intent-to-use filings exist, but a name already on sold merchandise is stronger ground than one that isn't.
Even a strong name fails if it collides with a senior mark in the same class. We screen for likely conflicts so the verdict reflects the field, not just your mark.
Three sources. Guidance, interpretation, and a final ruling. Cited verbatim.
Two checks run alongside the factor analysis — one before, one during. Here's what each is, and what it isn't.
The image is checked against known characters and recognizable protected IP. If something protected is detected, the assessment halts with a flag — the five factors never run.
We evaluate distinctiveness and flag likely conflicts with existing marks. This is a probabilistic assessment — it shapes the verdict, not a filing.
Three possible outcomes on the copyright track. Each comes with a specific explanation and — where applicable — the tools to strengthen your position and re-run the assessment.
Strong evidence of human authorship across most factors. Your work is well-positioned to register; the filing path is clear.
Some factors are strong, others thin. You can register, but the claim is narrower than it could be — strengthening weak factors before filing changes what's protected.
Current evidence of human authorship is too thin for registration. This is a starting point, not a verdict — there are specific ways to make the work yours.
Copyright law as applied to AI-generated work is actively evolving. This assessment reflects U.S. Copyright Office guidance, its January 2025 Part 2 Report, and federal court rulings through early 2026. It is a tool for self-assessment before you file — not a substitute for counsel, and not a guarantee of registration.