With the latest explosion of Artificial Intelligence (“AI“) generated content there comes the inevitable question of how to protect that content. There has been a lot of confusion over who owns the copyrights in the image. Does the AI generator, be it ChatGPT[1], DALL-E 2[2], Midjourney[3], or Bard[4] own it? Does the person who wrote the prompt; that is the current user own the work? Or maybe the owners are the artists whose work was used to train AI technology? Because AI generated content is so new, we do not have any guidance from the courts.
To provide some clarity about who owns the underlying rights, given prior court decisions and how to register works that are created with the aid of AI content generators the US Copyright Office issued a guidance[5] on the subject.
Most importantly for copyright protections to attach the author must be a human. This means that no AI generator has the legal ability to hold the copyright to a work. This is not a new concept. It has been around since 1884 when the concept of machine ownership was introduced in the seminal copyright case, Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53 (1884)[6]. At that time the debate was not over AI technology but another revolutionary technology, photography. The Supreme Court said that a camera is just a tool, just like a paint brush. With that same methodology, the Copyright Office is claiming that AI content generators are just tools, like cameras. They are not human and thus their works are excluded from protections.
This does not mean that when AI is included in work it is not eligible for protection. There is a test that looks to see if the work is basically one of human authorship with the device merely being an assistant, or whether the traditional elements of authorship were conceived and executed by a machine. One way this occurs is through a required disclaimer, indicating what portion of the work was created by a person, and what portion by the AI content generator. If the work is a novel that was written by AI content generators based on a single sentence prompt from the “author;” then it’s more likely than not that the copyright registration will be denied based on the lack of human authorship. This is because the AI will have made the “artistic” determinations about the novel, as opposed to the actual human author. This is not the same as having a novel written by a human and then using AI generated images in it, or AI checking your novel for grammatical or spelling mistakes. In that case, the novel will be eligible for copyright protections, but not the AI generated images.
If you have a mixture of human and AI created content, you must disclaim the portion of the work created by the AI. This can be done in the Limitations of Claim section of the Standard Application. Or, you can specify what sections you and the AI created in the Author Created field. Or, if the AI generated content is de minimis, super small, then there is no need to disclaim it. Please note that you do not need to list the AI content generator as a co-author. This would be the equivalent of list Adobe Photoshop on images that have been touched up.
If you have any questions about registering your work that was created in part by AI content generation, please feel free to reach out to me at adan@apexlg.com or if you want additional information, please visit our website. One final note, this blog is for educational purposes and does not constitute legal advice. This article, or contacting the Apex Law Group, does not form an attorney client relationship.
[1] https://openai.com/blog/chatgpt
[2] https://openai.com/research/dall-e
[3] https://www.midjourney.com/home/?callbackUrl=%2Fapp%2F
[5] https://copyright.gov/ai/ai_policy_guidance.pdf?loclr=twcop
[6] https://tile.loc.gov/storage-services/service/ll/usrep/usrep111/usrep111053/usrep111053.pdf