Game studios using Midjourney and other AI image generators for commercial projects may not be able to legally protect those assets, according to new guidance from the U.S. Copyright Office. The foundational rule is stark: copyright protects only works of human authorship, and raw AI output generated without substantial human creative contribution cannot be registered or defended against copying.

As studios rapidly integrate AI tools into their asset pipelines, the legal stakes are becoming clearer. Concept artists generate background plates in Midjourney, composers prototype tracks in AI music tools, and engineers scaffold dialogue systems with AI-generated code. The speed and cost savings are real. But the copyright questions are not going away, and the answers may be different from what many teams assume.

What Does the Copyright Office Actually Protect?

On January 29, 2025, the U.S. Copyright Office released Part 2 of its Report on Copyright and Artificial Intelligence, concluding that "the outputs of generative AI can be protected by copyright only where a human author has determined sufficient expressive elements". This distinction matters enormously for game studios deciding what to register and how to structure their creative workflows.

The spectrum of protectability breaks down into three categories:

• Raw AI Output: An image generated by Midjourney from a text prompt with no further human modification is not copyrightable, meaning anyone can copy it without legal consequence.
• Human-Arranged AI Compositions: If a human selects, arranges, and modifies AI-generated elements into a larger work, such as a concept art sheet where an artist composites AI-generated images into a hand-painted scene, the human's arrangement and modifications may be protectable even if the underlying AI elements are not.
• Human-Authored Works With AI Assistance: If a composer writes an original melody and uses AI to generate an orchestral arrangement that the composer then substantially revises, the human-authored elements are fully protectable.

The line between partial and full protection is where most studios will operate, and it is also where the legal uncertainty is greatest. The Copyright Office has promised ongoing registration guidance, but for now, the safest approach is to assume that raw AI output is unprotectable and to document the human creative contributions that go into every asset a studio intends to commercialize.

Why Did the Zarya of the Dawn Case Change Everything?

The most instructive precedent for game studios comes from the Copyright Office's decision on "Zarya of the Dawn," a comic book created by Kris Kashtanova using Midjourney-generated images. Kashtanova initially received a copyright registration for the full work, but after the Copyright Office learned that the images were AI-generated, it cancelled the original registration and issued a new one that covered only the text and the arrangement of images and text, not the AI-generated images themselves.

The Copyright Office's letter explained that "the images in the Work that were generated by the Midjourney technology are not the product of human authorship." Even though Kashtanova argued she had "guided" the structure and content of each image through prompts, the Office concluded that "it was Midjourney, not Kashtanova, that originated the traditional elements of authorship in the images."

This case established three critical practical takeaways for game studios:

• Prompting Is Not Authorship: Writing detailed prompts does not make you the author of the AI-generated output. The Copyright Office was explicit that the human's role in guiding the AI did not constitute the creative control necessary for authorship.
• Arrangement and Selection Can Be Protected: While individual AI-generated images were not protectable, Kashtanova's selection, coordination, and arrangement of those images within the comic was protected. This means a game's level design, UI layout, and asset composition may be protectable even if individual AI-generated assets are not.
• Disclosure Matters: The original registration was cancelled because Kashtanova did not disclose that the images were AI-generated. When registering a copyright, studios must disclose AI-generated content, or risk cancellation of the registration.

How Should Studios Protect Their AI-Generated Assets?

Given the Copyright Office's guidance and the Zarya precedent, game studios can adopt a practical framework for what they can and cannot register. The key is understanding which elements of a project qualify for protection and which do not.

• Human-Written Game Code: Code written by engineers, even if they used AI tools like GitHub Copilot for auto-completion, is generally registerable as long as the human engineer made the substantive creative and technical decisions. The Copyright Office acknowledges that AI can assist in the creation process without barring copyrightability.
• Human-Written Narrative and Dialogue: If a narrative designer writes NPC dialogue and uses AI only for brainstorming or proofreading, the written dialogue is human-authored and registerable.
• Human-Composed Music With AI Arrangement: If a composer writes the melody, harmony, and lyrics and uses AI to generate an orchestral backing track that the composer then substantially revises, the human-authored elements are registerable.
• Level Design and Asset Arrangement: The selection, coordination, and arrangement of assets within a game, even if some individual assets are AI-generated, may be registerable under the Zarya precedent.
• Raw Midjourney Concept Art: An image generated by Midjourney with no further human modification cannot be registered and offers no legal protection against copying.

The practical implication is clear: studios cannot list an AI system as the author of any asset in a copyright registration. When filing a registration, the author must be a human, and that human must have made a sufficient creative contribution to the work. Listing "Midjourney" or another AI tool as a co-author is not a viable legal strategy.

What About Compliance With New Detection Laws?

Beyond copyright concerns, studios also face new transparency requirements. California enacted a state artificial intelligence transparency law that went into effect on August 2, 2026, requiring AI companies with over one million users to mark their photo, audio, and video content with metadata and to make detection tools publicly available. A similar law went into effect in the European Union.

An independent review of 13 major companies that generate AI images and audio found that only 7, with an eighth identified by the reviewing organizations, published the legally required tools to help users determine whether content was made or edited with their technology. The authors found that HeyGen, Midjourney, Mistral, Synthesia, and xAI did not have tools available at the time of the review. Midjourney did not respond to requests for comment on its compliance status.

The California law aims to make AI content more identifiable, but the review also found that just because a detection tool exists does not mean it can reliably identify whether content was AI-generated. Most tools could identify their own unedited content, but only Google and OpenAI could identify their own output even after it was edited. Only Adobe and Microsoft could identify another tool's unedited content more than half of the time.

For studios operating in California or selling to California-based users, understanding these compliance requirements alongside copyright protections is essential. The combination of copyright uncertainty and detection tool requirements means that studios must be intentional about documenting human creative contributions and understanding the legal status of every asset they commercialize.