Uncertain Copyright and Patent Protections in the AI Age
Editor's note: Authored by Adam Gershenson, Angela Dunning, Kristin Leavy and Dustin Knight, this article was originally published in Law360.
Advances in artificial intelligence technology have enabled the generation of new inventions, text, images, videos, and other content with unprecedented speed and ease.
They also give rise to a crucial question: How do intellectual property laws protect what is created using AI?
For both copyright and patent law, the answer is evolving.
Recent decisions and pronouncements demonstrate that AI cannot be a wholesale substitute for human intelligence, skill and creativity. But that is not to say that all AI-assisted content is currently barred from protection in the U.S.
The Patent Act requires inventors to be natural persons.
Last year, in Thaler v. Vidal in the U.S. Court of Appeals for the Federal Circuit, the U.S. Patent and Trademark Office rejected an application for a patent covering output generated from AI software that the applicant created, where the application identified the AI as its would-be inventor.
The USPTO determined that an inventor must be an individual, and that an individual must be a natural person, precluding an AI as the inventor. The Federal Circuit agreed, but sidestepped "whether inventions made by human beings with the assistance of AI are eligible for patent protection."
The U.S. Supreme Court denied certiorari, leaving intact the Federal Circuit's decision.
Since Thaler, the USPTO has explored the role of AI in inventorship and its implications on intellectual property policy through its AI and emerging technology partnership initiative, and by seeking public input in a February request for comment.
These inquiries explore the degree to which AI systems influence the innovation ecosystem, the ways in which reliance on AI systems differs from the use of other technical tools in invention ideation, and what patent protection should be available if an AI system contributes to an invention at the same level as a human under the doctrine of joint inventorship.
Looking ahead, AI assistance will be key to inventions on everything from software to medical devices, as will the contours of a regime that determines when such output should be recognized as a patentable invention, with the attendant exclusive right to use, make and sell such products.
The Copyright Act has repeatedly been interpreted to require human authorship.
Although authors may utilize technological tools to assist in creating copyrightable works, those works must be created by humans. In August, in Thaler v. Perlmutter, the U.S. District Court for the District of Columbia granted summary judgment to the U.S. Copyright Office and denied a cross-motion for summary judgment filed by plaintiff Stephen Thaler.
Thaler argued that the Copyright Office had erred when refusing copyright registration for a work "autonomously created by a computer algorithm running on a machine" Thaler had invented and dubbed the "creativity machine."
In denying registration, the Copyright Office reasoned that copyright extends only to works created by human authors and not works created entirely by an AI.
The court agreed with the Copyright Office that copyrightability requires "human involvement in, and ultimate creative control over, the work," and that "human creativity is the sine qua non at the core of copyrightability, even as that human creativity is channeled through new tools or into new media."
Works originating entirely from nonhumans are not protectable under copyright law, but human creativity coupled with technology may result in a work that is protectable under copyright, like photos taken by a human using a camera.
The court acknowledged that "we are approaching new frontiers in copyright" and that many open questions remain, including
how much human input is necessary to qualify the user of an AI system as an "author" of a generated work, the scope of the protection obtained over the resultant image, how to assess the originality of AI-generated works where the systems may have been trained on unknown pre-existing works, how copyright might best be used to incentivize creative works involving AI, and more.
Although the court in Thaler addressed the narrow issue of whether a work created by artificial intelligence without human creativity is protectable under copyright, the Copyright Office has been wrestling with the new frontier of how much human involvement is necessary for copyright registration.
According to the Copyright Office, as of June, it had received less than 100 claims of works known to incorporate content created using generative AI.
In its 2021 Compendium of U.S. Copyright Office Practices, the Copyright Office clarified that "to qualify as a work of 'authorship' a work must be created by a human being," and that it "will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author."
In order to further clarify its practices for examining and registering works that contain AI-generated materials, the Copyright Office issued copyright registration guidance for these works, which requires disclosure of the inclusion of unclaimable AI-generated content and the claimable human contributions in a work submitted for registration.
Applying these principles, in its registration decision regarding the comic book "Zarya of the Dawn," the Copyright Office denied protection for images created using the generative AI art platform Midjourney Inc.
The office indicated that rather than being a tool that the author "controlled and guided to reach her desired image, Midjourney generates images in an unpredictable way." Under this view, the prompts that a user provides are merely suggesting, rather than directing, what the content output will look like.
The Copyright Office allowed registration for the text and the selection and arrangement of images and text where the applicant, Kris Kashtanova, attested Kashtanova was solely responsible for those elements.
Kashtanova's second, still-pending application, points to a new doctrinal frontier. That application raises the issue of whether copyright law will cover as a derivative work an image, titled "Rose Enigma," that originated with a copyrighted drawing by a human author but was then iterated upon with an AI tool to add color and dimensionality but otherwise retain substantial similarity of the original expression.
The Copyright Office has yet to weigh in on this scenario, but a June webinar on registration guidance for works containing AI-generated content reiterated the office's position that AI-developed or modified works are not eligible for protection under copyright law, using an example of an illustration modified by AI:
[I]f you plan to use AI to make appreciable changes to a photograph or illustration that was created by a human author, you can register the human authored photograph or illustration without the AI modifications, even if the only version that you intend to publicly display or publicly distribute is the modified AI version. If AI is used to make only de minimis changes, then it does not matter whether you registered before or after, because in neither case would disclosure be required.
Protection only accrues to the human-authored components and not to any of the AI tool manipulations, but the resultant image or work may nevertheless be substantially similar to the original copyrighted one, such that it constitutes a protectable derivative that third parties may not copy without permission.
The Copyright Office's decision with respect to the registrability of "Rose Enigma" will determine this issue, and may also elucidate whether there may be circumstances in which authors exert sufficient control over an AI tool — e.g., by including their copyrighted material in detailed prompts — so that they can control what is generated, resulting in output protectable under copyright law.
Until Congress, courts or regulators take more definitive action, the landscape will feature rapidly advancing technology governed by outmoded rules that cannot keep pace.
Still, possibilities abound.
Both the USPTO and the Copyright Office have left open the possibility that AI-assisted works will be protectable under patent and copyright laws, and are investigating how to enable some form of intellectual property protection for AI-enabled innovation.
For example, where inventors and authors exert sufficient control over the AI to demonstrate user-directed output, protection may arise.
Although based on current guidance, not all use of generative AI technology will produce copyright-registerable content, the Copyright Office will consider whether the AI contributions are an author's "own original mental conception, to which [the author] gave visible form" in determining whether the human direction was sufficient.
The Copyright Office will make this determination on a case-by-case basis, and the "answer will depend on the circumstances, particularly how the AI tool operates and how it was used to create the final work."
The Copyright Office has acknowledged that, like everyone else, it is still learning about generative artificial intelligence.
As it has the chance to review more copyright applications and research the technologies used to create works subject to those applications, the Copyright Office expects to provide more guidance on authorship and the scope of protection for works generated in whole or in part with AI tools. Control and direction are likely to be key components in its analysis.
Similarly, in the patent realm, we might expect more protection to arise where a human inventor enjoyed assistance from AI, but nonetheless made the primary novel contributions to the invention.
Given the uncertainties in the law, for now, would-be authors and inventors should closely document their personal contributions to their works and inventions, as contrasted with contributions made by AI technologies.
This is particularly important for authors, as those seeking to avail themselves of a copyright registration must also disclose the existence of AI-generated content or components to the Copyright Office, with the understanding that those components may not be protectable or claimable — or risk wholesale rejection of their copyright registrations.
Given the potentially evolving nature of these legal issues, authors and inventors should be prepared to identify how and with which technologies AI contributed to their works.
If an approach more favorable to authors and inventors who use AI is adopted by the Patent Office or Copyright Office and permits protection where a human inventor was assisted by AI, this documentation will be critical to support intellectual property protection.
AI itself seems to expect a balanced approach.
When asked recently whether AI-generated works should be protected by copyright or patent law, ChatGPT opined that ultimately, "finding a balance between encouraging innovation and creativity while respecting ethical considerations is crucial."
The coming years will test how well our legal regimes strike that balance.