DALL-E image based on the prompt "robot signing his name on a painting" (image generated by Valentina Di Liscia/Hyperallergic using DALL-E)

The United States Copyright Office recently produced a statement of policy indicating that some artworks generated using artificial intelligence are now eligible for copyright registration on a case-by-case basis. This should go well!

Effective March 16, the Copyright Office’s statement of policy indicates that copyright applicants are permitted to submit AI-assisted works (across literature and visual arts) for protection under copyright law, and that the works will be evaluated for evidence of “human authorship.” The Office made a comparison between AI art and photography, citing the 1884 Supreme Court decision to extend copyright protections to photographs against the will of the Congress in Burrow-Giles Lithography Co. v. Sarony.

The Supreme Court decided that a photograph is not just a mechanical process, but an authored work based on the photographer’s decisions in curating the backdrop and subject’s clothing.

In the realm of generative works, the Office asks applicants if the included AI elements are the result of “mechanical reproduction” or of an author’s “own original mental conception, to which [the author] gave visible form.” (Walter Benjamin: 💅) To mark the difference, the policy distinguishes between human artists developing AI work strictly through submitting prompts as instructions, and human artists selecting from and reimagining AI generations in a “sufficiently creative way.” However, the policy states that in the latter case, only the “human-authored” elements of the work would be copyrighted independent of the AI contributions.

The Office cites the example of a 2018 work generated autonomously by an unattended computer algorithm that was submitted for copyright protection and ultimately rejected as it was developed “without any creative contribution from a human actor.” On the other hand, a graphic novel with human-written text and Midjourney-generated imagery was granted copyright protection as a whole, but the individual images were omitted from the approval as they were not considered works of human authorship.

In order to get submissions approved for copyright registration, the office states that “applicants have a duty to disclose the inclusion of AI-generated content in a work,” and should provide an explanation outlining the human-authored elements that were contributed versus the AI-assisted elements.

In the eyes of the Copyright Office, the human hand has the upper hand for now, which should be a relief for practicing artists concerned about AI encroaching on the commissions and freelance market.

Rhea Nayyar (she/her) is a New York-based teaching artist who is passionate about elevating minority perspectives within the academic and editorial spheres of the art world. Rhea received her BFA in Visual...