Copyright Office’s new policy for AI artworks: authors have a “duty to disclose” and “explicitly exclude AI-generated content” from copyright registration.

The U.S. Copyright Office’s new guidance for copyright registration of works containing more than de minimis content generated AI is controversial. Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, 88 Fed. Reg. 16,190 (Mar. 16, 2023) (to be codified at 37 C.F.R. § 202). In this article, we take a look at the most controversial aspects of it.

Applicants have a “duty to disclose (1) the inclusion of AI-generated content in a work” and (2) “provide a brief explanation of the human author’s contributions to the work.”

So, what is the source of this duty to disclose AI-content in an application for registration? While patent applications with the USPTO have long required a duty of candor and good faith in dealing with the USPTO, including a duty to disclose prior art and other information related to patentability known by the applicant, the phrase “duty to disclose” may be comparatively less often used for copyright registrations, which, for works whose origin is the United States, is a prerequisite to filing a copyright lawsuit (but not a condition of acquiring a copyright, which is automatic upon fixation of an original work of authorship).

Despite the apparent infrequent use of a copyright owner’s “duty to disclose,” the Copyright Act does require, in Section 409, that an application “shall include” “(10) any other information regarded by the Register of Copyrights as bearing upon the preparation or identification of the work or the existence, ownership, or duration of the copyright.” The Copyright Office’s new guidance now requires the disclosure of any content that is more than de minimis that was generated by AI, plus the explicit exclusion of that AI content from the claim of registration. This duty to disclose would apply to ChatGPT text-to-image generators, including DALL-E, Midjourney, and Stable Diffusion. The Copyright Office has already rejected Kris Kashtanova’s application to register images generated through Midjourney. (For more on that case, click here.)

[1] DUTY TO DISCLOSE AI generated content: “Individuals who use AI technology in creating a work may claim copyright protection for their own contributions to that work. They must use the Standard Application,39 and in it identify the author(s) and provide a brief statement in the ‘‘Author Created’’ field that describes the authorship that was contributed by a human.”

[2] DUTY TO EXCLUDE AI generated content: “AI-generated content that is more than de minimis should be explicitly excluded from the application. This may be done in the ‘‘Limitation of the Claim’’ section in the ‘‘Other’’ field, under the ‘‘Material Excluded’’ heading. Applicants should provide a brief description of the AI-generated content, such as by entering ‘‘[description of content] generated by artificial intelligence.’’ Applicants may also provide additional information in the ‘‘Note to CO’’ field in the Standard Application.”

[3] “Applicants who are unsure of how to fill out the application may simply provide a general statement that a work contains AI-generated material. The Office will contact the applicant when the claim is reviewed and determine how to proceed. In some cases, the use of an AI tool will not raise questions about human authorship, and the Office will explain that nothing needs to be disclaimed on the application.”

What happens if an applicant fails to disclose and exclude AI-generated content in the copyright registration application?

Under the Supreme Court’s interpretation of Section 411(b) of the Copyright Act, a copyright registration can still be valid–a copyright lawsuit maintained based on the registration–if the applicant/copyright owner lacked knowledge that its application was inaccurate in terms of its understanding of facts or law. Unicolors, Inc. v. H&M Hennes & Mauritz, L.P., 142 S. Ct. 941 (2022).

In other words, if an applicant had no knowledge of the Copyright Office’s AI-content disclosure requirement, and was not willfully blind, a copyright registration could be valid and lawsuit maintained based on it, even though the applicant had failed to disclose that AI was involved in the creation of the work of authorship.

Conversely, if the applicant had such knowledge and failed to disclose the information as required by the Copyright Office’s guidance, that failure to disclose would make the registration itself vulnerable to attack as invalid. Section 411(b) states: “In any case in which inaccurate information described under paragraph (1) is alleged, the court shall request the Register of Copyrights to advise the court whether the inaccurate information, if known, would have caused the Register of Copyrights to refuse registration.” If the registration is invalid, the copyright owner cannot maintain an infringement lawsuit in court. An invalid registration doesn’t mean the copyright is forfeited. Instead, it means that the copyright owner can’t bring a lawsuit to enforce their copyright until it secures a valid registration for the work with the correct information.

Can an AI artist follow the Copyright Office’s disclosure and exclusion requirements, but later challenge the Office’s denial of copyrightability of content generated from text-to-image prompts? Or is the challenge waived?

One lurking question is whether an AI artist who follows the Copyright Office’s new disclosure and exclusion requirements can later challenge, in court, the Copyright Office’s “human authorship” approach that treats content generated with text-to-image prompts using ChatGPT programs as uncopyrightable. In other words, does an AI artist’s “explicit” exclusion of such AI-generated content in an application for copyright registration constitute a waiver of any later challenge to the Copyright Office’s legal position in denying the copyrightability of content generated by text-to-image generators?

Great question. Offhand, I’m not sure. One can insert a notation attempting to preserve a legal challenge to the Copyright Office’s view of “human authorship” in the as additional information in the ‘‘Note to CO’’ field in the Standard Application. But I am not sure this is effective to avoid any waiver issue and will investigate.

Copyright registration is not required for works of foreign origin to file a copyright lawsuit, although statutory damages are only available to registered copyrights

It’s worth noting that copyright registration is a prerequisite to filing a copyright infringement action only for “United States works“–works first published here or, for unpublished works, if the author is a U.S. national or U.S. resident. See Section 411.

That means that registration isn’t required for works first published in foreign countries. Likewise, for unpublished works, foreigners who are not US residents don’t have to register their copyrights before filing a copyright infringement lawsuit in the United States. See Section 101 definition of “United States work.” However, any author who hasn’t filed registration of US copyright for a work is not eligible to seek statutory damages or attorney’s fees as a prevailing party in an infringement lawsuit involving that work. See Section 412.

2 responses to “Copyright Office’s new policy for AI artworks: authors have a “duty to disclose” and “explicitly exclude AI-generated content” from copyright registration.”

Leave a Reply

%d bloggers like this: