The US Copyright Office denies protection for an AI-made image that won an art competition, despite Adobe Photoshop alterations and 624+ text prompt revisions
thus nothing could be granted to him. [image] @glynntarrant : It's good to see Jason is still struggling with trying to copyright this AI generated image. Also, since he was and still is unwilling to disclaim the AI generated image, he's going to be stuck in this legal loop for a while. @franklingraves : The @CopyrightOffice has issued a denial for a Second Request to reconsider its refusal to register an AI-generated work... that one that won the art competition 😳 In short: Jason Allen failed to meet the de minimis standard. #copyright #AI #GenerativeAI [image] @neermcd : Yesterday the U.S. Copyright Office affirmed a refusal denying registration of the below image. USCO stated the work is more than a de minimis amount of content generated by AI, and therefore, must be disclaimed. [image] @coreybrickley : sorry boys, paint overs won't cut it [image] Reid Southen / @rahll : Excellent thread from an IP lawyer. The US copyright office has again refused to allow copyright registration to that AI generated painting that won an art competition. AI advocates keep saying how artists are losing the battle, adapt or die, yet here we are with another win... @franklingraves : The takeaway: Using Midjourney? Be prepared to disclaim when registering a work, and then be left questioning what you actually own and are able to protect 😅 Using any other AI tool? Be prepared to disclaim, and also question what's protectable. Tom Coates / @tomcoates : @sampullara “Real artists” were the ones that suffered when he won that prize https://www.nytimes.com/... Wagner James Au / @slhamlet : I feel like the “Generative AI will totally transform Hollywood and all media!” hype is sorta undercut by the small matter that it can't be copyrighted? https://www.reuters.com/... Forums: Hacker News : US rejects AI copyright for famous state fair-winning Midjourney art Msmash / Slashdot : US Copyright Office Denies Protection for Another AI-Created Image Ars OpenForum : US rejects AI copyright for famous state fair-winning Midjourney art See also Mediagazer
Context & Ripple Effects
The decision applies the Copyright Office’s earlier human-authorship line: protection for AI-assisted work turns on whether the protected expression reflects the creator’s own mental conception, not simply the amount of prompting or post-production.
It also narrows the apparent ambiguity created when a Midjourney-based comic received a copyright registration with AI-generated artwork, while following the Office’s earlier rejection of an AI-created artwork lacking human authorship.
First-order effects
- Jason Allen cannot obtain registration for the image as submitted, because the Office did not treat the Midjourney-generated expression as sufficiently attributable to him despite the prompt iterations and Photoshop work.
- Creators using image generators have a clearer immediate limit: edits and instructions do not by themselves secure protection for the generated image unless the human-authored contribution is separable and meaningful.
Second-order effects
- Creative professionals and platforms will have stronger incentives to document human-made elements and to structure workflows around clearly identifiable edits, rather than rely on prompt volume as evidence of authorship.
- For buyers and licensors, exclusive-rights claims over largely AI-generated visuals become harder to rely on, potentially shifting value toward human-authored components and contractual rights.
Third-order effects
- If this approach persists, generative-image markets may split between works with demonstrable human authorship that can be registered and abundant AI output that is harder to monopolize through copyright.
- The decision reinforces a policy boundary around human authorship while leaving continuing pressure on courts and lawmakers to address ownership, training data, and infringement disputes involving generative systems.
The trend: Generative-AI commercialization is being shaped by a growing divide between using AI as a tool within human authorship and seeking copyright in AI-generated output itself.