A software glitch on January 23, 2023, in the United States Copyright Office’s record-storing software incorrectly revoked the copyright registration of a graphic comic book produced by artificial intelligence.[1] The comic book titled Zarya Of The Dawn, which featured images created by Midjourney and filed by artist and AI consultant Kristina Kashtanova, had its application approved beforehand by the Copyright Office (Office) on September 15, 2022.[2] In what could be determined as the first acceptance of an AI-generated, copyrighted work in the United States, the Midjourney software is responsible for the image creation by its own text-to-image tool.
Although the glitch was quickly corrected, the situation raises a few important questions about this intersection between AI and copyrights within the United States. This article will address:
- The “human involvement” factor as inextricably linked to U.S. copyright law;
- Pending litigation from a rejected AI-based copyright application; and
- The Office’s manipulation of AI as a business tool in its creation of a small claim tribunal.
By addressing these few significant legal developments concerning AI and the Office, which undoubtedly could be expanded given the wealth of recent filings and scholarship, this topic should lose some of its novelty.
Circling back to Kashtanova’s filing, the decisions of the Office seem to emulate their roundabout approach to issuing guidance and regulations pertaining to AI-creations. On October 28, 2022, a month after Kashtanova’s filing, the Office cited that her registration might be canceled.[3] Poignantly, the Office asked Kashtanova to provide further details of her “process to show that there was substantial human involvement in the process of creation of this graphic novel.”[4]
This request echoes the Office’s succinct justification as provided in recent determinations against AI-created works seeking copyright registration. All rejected applications have seemingly centered on the degree of “human involvement” and their lack thereof as attributed to the final product (“creation”).[5] More specifically, the Office has interpreted the Copyright Act to require human authorship since copyright is limited to “original intellectual conceptions of the author.”[6] However, does this notion of “author” in the Act apply to only humans?
As the field of AI continues to grow nationally, internationally and across disciplines, press and researchers have questioned whether the Office can provide express limitations defining what creations using AI may be accepted for copyright registration. In an effort to answer one press source, the Office replied it “will not knowingly grant registration to a work that was claimed to have been created solely by machine with artificial intelligence.”[7]
Although this statement proves helpful should a work be exclusively sourced from AI, it affords no direction or sliding scale for creations that may be heavily dependent on AI. Would a work meeting a threshold of 50%, 75% or 90% of AI-involvement be accepted for registration by the Office? The Office and U.S. legislators should consider immediately quantifying the degree of “human involvement” necessary to impart copyright registration.
A rudimentary sliding scale would assist potential applicants, provide direction for the development of future AI configurations, and save the time and money afforded to the registration process on both ends – applicant and Office.
Furthermore, the question arises as to the veracity of the application itself: did the AI receive proper credit for its contribution? Moreover, how does the Office validate “human involvement”? A section titled “Limitation of Claim,” which is included as part of any copyright registration, may offer an answer along with more possibilities. The “Limitation of Claim” field allows applicants to disclaim credit for uncreditable material, which includes the following: derivative works; compilations; “material that was previously published; material that was previously registered; material that is in the public domain; and/or material that is owned by an individual or legal entity other than the claimant who is named in the application.”[8] This section most likely serves as the most appropriate place to disclaim any AI-involvement from the particular application. For example, Kashtanova most likely disclaimed Midjourney’s AI-created images in this area of her registration.
However, there is current debate as to whether any “human” element should be associated with copyright law. On January 10, 2023, computer scientist Stephen Thaler asked the U.S. District Court for the District of Columbia to overturn an Office decision that stated creative works must be made by humans to receive copyright protection.[9] Thaler’s recent filing noted that the Office’s ruling was based on outdated precedent as the U.S. Copyright Act “does not restrict copyright to human-made works, nor does any case law.”[10] Although Thaler’s argument weighs heavily on the blackletter interpretation of copyright law, the Office hopes to utilize specific linguistic choices found in cases and decisions to sway the court’s decision.[11] Since this matter is still pending, we should expect 2023 to be ripe with new filings, determinations, and further questions regarding language construction.
Speaking of semantic practices, the terms and conditions of AI-generation tools, such as Midjourney, may place further restrictions on the type of license granted to users of the software. Depending on their express terms, users such as Kashtanova may be prevented from seeking ownership or rights associated with any AI creation.
With the rise in matters concerning copyright, Congress passed the Copyright Alternative in Small-Claims Enforcement (CASE Act) in 2020, which established a Copyright Claims Board (CCB) to resolve disputes outside the federal court system.[12] Three Copyright Claim Officers (CCOs), who have deep expertise in copyright law, render determinations via an online system rather than judges.
The CCB offers a variety of claims, including: a party accused of infringement can file a claim seeking a ruling that its actions do not infringe; a claim for misrepresentation in connection with a Digital Millennium Copyright Act (DMCA) notification; or a counter notification seeking to replace removed or disabled material.[13] The respondent, whom the claim is made against, maintains the ability to opt-out of the proceedings so long as it occurs within 60 days of notice.[14] The claimant may then choose to file suit against the respondent in federal court. If no opt-out occurs before the 60-day deadline, the proceeding moves forward in the CCB with or without their participation and the respondent will be bound by the determination.[15]
Since a federal lawsuit cannot commence until copyright registration is complete, the CCB offers a quicker and less expensive method of obtaining copyright determinations. Many business and creators have initiated claims using this cost-effective, timely system that should only grow in the future.[16] Although no final decisions have yet to be rendered, the Office’s reliance on this online tribunal captures the ability of artificial intelligence to assist in streamlining legal processes. The virtual CCB, Office correspondence, and legal documentation within the platform are all technological advancements akin to the usage of AI in the creation of copyrighted works. Without technology, the CCB would not exist as an alternative forum. Applying the standards set by the Office itself, how much “human involvement” is associated with the CCB’s business practices?
Considering the peculiar technological glitch and initial possibility of cancellation, Kashtanova’s graphic comic book functions as the perfect metaphor for the convergence of AI and copyright in the U.S. Although it appears as a novel issue at first glance, its underpinnings are well documented as the Office and legislative branch attempt to adapt to our current world. Just as Thaler articulated, outdated precedent and semantic preferences require updating as society progresses. Our culture has quickly adapted to different pronouns not linked to gender or sexual orientation and our need for “human involvement” has been tested during the height of the pandemic. To be required to expressly identify and disassociate what is “non-human” under the law in order to obtain copyright protection seems antiquated. Even the Office relies heavily on its “non-human” creations.
[1] Katyanna Quach, Software Glitch Revokes Copyright Protection for AI-Generated Comic Book, The Register (Jan. 25, 2023), https://www.theregister.com/2023/01/25/glitch_us_copyright_office_ai/.
[2] George He, AI – Is it Art, Yet?, JOLT Digest – Harvard Law School (Oct. 20, 2022), https://jolt.law.harvard.edu/digest/ai-is-it-art-yet.
[3] Franklin Graves, U.S. Copyright Office Backtracks on Registration of Partially AI-Generated Work, IP Watchdog (Nov. 1, 2022, 12:15 PM), https://ipwatchdog.com/2022/11/01/us-copyright-office-backtracks-registration-partially-ai-generated-work/id=152451/.
[4] Id.
[5] James Vincent, The Scary Truth about AI Copyright is Nobody Knows What Will Happen Next, The Verge (Nov. 15, 2022, 10:00 AM), https://www.theverge.com/23444685/generative-ai-copyright-infringement-legal-fair-use-training-data.
[6] Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 58 (1884).
[7] Riddhi Setty & Isaiah Poritz, ‘Wild West’ of Generative AI Poses Novel Copyright Questions, Bloomberg Law (Nov. 18, 2022, 11:00 AM), https://news.bloomberglaw.com/ip-law/wild-west-of-generative-ai-raises-novel-copyright-questions.
[8] U.S. Copyright Office, Help: Limitation of Claim, Copyright.gov, https://www.copyright.gov/eco/help-limitation.html (last visited Feb. 1, 2023).
[9] Thaler v. Perlmutter, U.S. District Court for the District of Columbia, No. 1:22-cv-01564.
[10]Id.
[11] Dennis Crouch, An Update on AI Inventorship and Authorship Cases, Patently-O (Jan. 24, 2023), https://patentlyo.com/patent/2023/01/update-inventorship-authorship.html.
[12] CASE Act of 2019, H.R.2426, 116th Cong. (2019), https://www.congress.gov/bill/116th-congress/house-bill/2426/text.
[13] Id.
[14] Id.
[15] Id.
[16] The CCB officially launched June 16, 2022. Copyright Claims Board, Frequently Asked Questions, CCB.gov, https://ccb.gov/faq/ (last visited Feb. 1, 2023).
