AI Prompts and Copyright: Developments in Music Industry
Introduction
An intriguing decision regarding a copyright registration matter involving AI-generated work was pronounced by the US Copyright Office (USCO) on September 5, 2023, as elaborated here. Earlier, in the Thaler case, an application naming the AI-system as author had been rejected for registering an AI-generated work.
The present issue pertained to whether an AI system can be considered an author- which has now been answered negative due to human authorship being deemed indispensable according to the ruling of USCO. In contrast to the Thaler situation, Jason Allen took steps to secure copyright registration for his piece of AI-generated art titled “Théâtre D’opéra Spatial,” listing himself as its creator. The Midjourney image AI system was responsible for producing this remarkable work that has captured nationwide interest by winning top honors at Colorado’s State Fair fine arts contest in 2022—marking it a groundbreaking achievement in the realm of art generated through artificial intelligence. [i]A couple of days ago, the USCO declined to register this piece for reasons that were dissimilar from those in the Thaler case since authorship wasn’t a concern. The problem with this submission was its failure to conform to the de minimis standard and disclose/disclaim AI-generated segments of it. It turned out that these portions surpassed what is considered negligible, prompting denial of registration as the applicant refused disavowing them despite wanting approval for their entire composition.
The copyright application submitted by the candidate failed to name Midjourney as the author and neglected to state that it was created using AI. Nevertheless, because of its national award-winning status, the Copyright Office recognized that it had been generated by artificial intelligence. In addition, the applicant asserted that he played a vital role in crafting the image. He offered numerous text prompts and even refined them to enhance their quality using software capable of eliminating defects while upgrading details. Notably, regarding these written cues, he maintained influencing the overall mood portrayed by the picture through his dictation of its tone.
Despite the applicant providing numerous prompts for the creation of the image, the USCO was still skeptical and referred to its AI Registration Guidelines (Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence). [ii]The guidelines stipulate that a work’s “traditional elements of authorship” must be produced by a human rather than a machine. In this particular case with Midjourney system processing Mr. Allen’s prompts how creative those inputs did not matter since output was based on interpretation from Midjourney using training data set it processed before ultimately creating an outcome distinctively determined technology instead of being created mainly through human efforts as these machines cannot understand instructions like humans would do; therefore breaking copyright law rules due to more than de minimis use outside what traditional automatons can perform gave authority to rule against approving under such circumstances by U.S Copyright Office(USCO).
Indian Context
The Indian Copyright Office is facing confusion as AI-generated art gains momentum. Such works are created solely by artificial intelligence without any creative input from humans, resulting in them being classified as “computer-generated works” under the 1995 Indian Copyright Act. The term ‘computer-generated work’ lacks definition and should be interpreted literally. Debates have arisen regarding human authorship requirements; some argue that it necessitates human involvement due to copyright duration being based on natural age. Currently, there are no Indian precedents illustrating whether computers/software can hold authorship nor official guidance from the Indian Copyright Office declaring only humans eligible for such recognition.
Furthermore, the India Copyright Office and Indian courts have not addressed how to interpret the phrase “the person who causes the work to be created.” It is unclear whether it refers to whoever creates the AI system or whoever supplies prompts (which are necessary for creating a work). However, following USCO’s reasoning, one could argue that even creative human inputs/prompts do not directly instruct/cause the creation of final works. Instead, such inputs only influence an AI system which then produces output based on its training data set and interpretation of supplied prompts. The applications of AI-generated images span numerous fields including digital art, training simulations, marketing initiatives personalized avatars, and 3D models. Due to its diverse usage, how copyright disputes regarding this content are addressed by courts will substantially impact those looking to safeguard their AI-generated creations.
The Indian Copyright Office is facing uncertainty in processing applications involving artificial intelligence. In 2020, an application claiming sole authorship for an artwork by the AI RAGHAV was rejected, while a subsequent application naming both a natural person and RAGHAV as co-authors received registration without clear reasoning. However, it seems that this approval may have been granted mistakenly since the Copyright Office issued a withdrawal notice almost one year later urging the applicant to clarify the legal status of the Raghav Artificial Intelligence Painting App themselves rather than relying on copyright office decision-making.
The applicant, who is a co-author and human, contends that revoking a granted copyright registration is not as straightforward as withdrawing it. It requires filing for rectification with the court to cancel the registration entirely. Although this notice was released in November 2021, there has been no update on the status of the proceeding yet. However, according to records from the Copyright Office’s website, this application remains registered at present. Although this may be old information, it remains relevant since AI-generated artwork continues to gain momentum rather than fade away. As a result, we ought to continue contemplating its implications.
The application can serve as a precedent for future ones until it is officially removed from the register or amended following regulations. Despite its current view on authorship requirements, the Copyright Office has limited options to change course and may face legal challenges if subsequent applications are unfairly rejected. Resolving this issue through litigation could prove costly and time-consuming for the office – an outcome that might have been preventable had it not reversed its initial ruling so abruptly.
It is alarming to observe copyright registrations being granted without due consideration, particularly as High Courts are now treating them as impregnable licenses rather than mere administrative results. The Gujarat High Court’s recent ruling that Copyright Office registration certificates offer protection against infringement immunity, which has been clarified to be an incorrect interpretation, only adds further concern. One issue is the absence of a legal precedent on whether AI can be recognized as an author. A Parliamentary Standing Committee has proposed revisions to IPR laws and adopted a robust framework for extracting gains from AI, but the legislation remains ambiguous. On another note, it might be feasible to acquire copyright registration naming AI as co-author while using this safeguard against litigation alleging breach of copyright – creating a concerning contradiction.