In a swiftly changing **AI landscape** and amid rising concerns regarding the **impact of artificial intelligence** on artistic endeavors, the U.S. government has established a definitive position: without direct human involvement, works cannot receive legal copyright protection.

In its detailed **report on AI and copyright**, the U.S. Copyright Office has confirmed that unmodified outputs from generative AI technologies are **not eligible** for federal copyright protections. This conclusion is part of a long-term effort to tackle the legal challenges presented by the AI surge, including the question of whether the Copyright Clause of the U.S. Constitution permits protection for works generated by AI.

“The results produced by generative AI can only be copyright-protected if a human author has introduced sufficient expressive components,” the Copyright Office declared in the second part of its report, issued on January 29. “This applies to situations where a human-created work is apparent in the AI output, or where a human makes creative alterations or arrangements of the output. However, simply providing prompts does not suffice.”

### The Essence of Human Creativity

Works that utilize generative AI in their creative process or that include elements generated by AI can still be eligible for copyright, as long as they uphold the **”centrality of human creativity.”** Shira Perlmutter, the Register of Copyrights and Director of the U.S. Copyright Office, stressed that allowing protection for materials where expressive aspects are determined by a machine would contradict the constitutional intent of copyright law.

Consequently, content produced with tools like **Midjourney** or **OpenAI’s DALL-E 3** cannot be copyrighted by the creators, even if they provide sophisticated, original prompts. Prompts themselves, in addition to iterative versions of AI-generated material, are also not eligible for copyright protection.

### The Importance of Human Engagement

The report also includes criteria for evaluating the extent of human participation in the creation of AI-assisted works, including the employment of computer-generated imagery in film production. A significant factor in this determination was the **unpredictability of generative AI outputs**, which can differ greatly even with identical or similar prompts.

“While entering prompts into a generative AI system may resemble instructing a commissioned artist, essential differences exist,” the report clarifies. “In human-to-human collaborations, the hiring party can monitor, guide, and fully comprehend the contributions of the commissioned artist. In contrast, the disparity between prompts and AI-generated outputs reveals that users do not possess control over how their ideas are converted into a fixed expression.”

### What Lies Ahead?

The third part of the report, anticipated later this year, will explore further the role of copyright in the training of AI models and the use of protected materials within generative AI tools.

For the moment, the U.S. Copyright Office has stated its position clearly: **human creativity is fundamental to copyright protection**, and works generated by AI, lacking significant human input, do not come under legal protections.