Can AI-Generated Output Be Protected Under Intellectual Property Law?
By Audrey Zhang Yang
Introduction
AI-generated output represents a groundbreaking integration of technology and creativity that increasingly challenges established norms in the legal world. Inevitably, it raises the question on whether law and policy on intellectual property protection should evolve and adapt to recognize this changing innovation trend. The Progress Clause of the Constitution gives Congress the power to “promote the Progress of Science…by securing for limited Times to Authors…the exclusive Right to their…Writing.”[i] Pursuant to this authorization, the Copyright Act extends copyright protection for “original works of authorship fixed in any tangible medium of expression.”[ii] The Copyright Act neither defined “authorship” not “works of authorship.”[iii] Traditionally, courts assigns authorship to individuals who create original works. However, determining authorship is more challenging in the case of artificial intelligence (AI). Some believe that since AI systems are tools programmed by humans, the programmers are entitled to authorship rights.[iv] Also, when someone instructs AI to solve a problem, that person might qualify as an investor if she formulates a problem in a manner that requires inventive skill.[v] However, laws on intellectual property, patent, and copyright were not originally passed with AI in mind. Therefore, there is no law specifically addressing AI-generated invention in any jurisdiction.
This paper first outlines recent cases where the courts refuse to recognize AI systems as copyright holders and patent inventors. It then analyzes the rationales behind the courts’ rejections of AI authorship and inventorship claims. Next, it presents arguments for and against extending IP protections to AI-generated works. Finally, it proposes potential regulatory responses to balance innovation incentives and the public interest. The complex and chaotic intersection of IP law, technology, and creativity reveals confusion, uncertainties, and challenges that society must address in the age of AI.
The Creativity Machine Copyright Dispute
In 2018, computer scientist Stephen Thaler decided to bring a series of test cases regarding the effect of AI on the current state of patent and copyright law in a wide range of jurisdictions including the US, the UK, Australia, South Africa, Canada, Germany, Switzerland, New Zealand Israel, Singapore, China, India, and Japan. In a website called “The Artificial Inventor Project”, Stephen Thaler and his international team of patent attorneys explicitly stated their goal included a series of pro bono legal test cases “seeking intellectual property rights for AI-generated output in the absence of a traditional human inventor or author. It is intended to promote dialogue about the social, economic, and legal impact of frontier technologies such as AI and to generate stakeholder guidance on the protectability of AI-generated output.”[vi]
Stephen Thaler listed his computer system as an artwork’s creator, arguing that copyright should be issued and transferred to him as the machine’s owner under common law property principle or the work-for-hire doctrine. In his copyright registration application to the U.S. Copyright Office (“USCO”) in 2018, Stephen Thaler identified the author as the Creativity Machine, which was a generative AI, and explained the work had been “autonomously created by a computer algorithm running on a machine” without any human authorship.[vii] He sought to claim the copyright of the “computer-generated work” himself “as a work-for-hire to the owner of the Creativity Machine.”[viii] In 2019, The USCO denied the application on the grounds that the work “lack[ed] the human authorship necessary to support a copyright claim.”[ix]
In 2022, Stephen Thaler filed a complaint in the U.S. District Court for the District of Columbia against the USCO and its director requesting the AI-generated artwork be registered because the Copyright Act provides protection to “original works of authorship.”[x] Thaler’s argument was based on his assertion that the Copyright Act does not explicitly define “author”, whose ordinary meaning could encompass AI.[xi]
The D.C. district court disagreed with Stephen Thaler’s definition of “author”. Judge Beryl Howell explained that an “author” is “an originator with the capacity for intellectual, creative, or artistic labor,” which is necessarily a human being.[xii] In her opinion, Judge Beryl Howell wrote, “Plaintiff can point to no case in which a court has recognized copyright in a work originating with a nonhuman…we are approaching new frontiers in copyright as artists put AI in their toolbox to be used in the generation of new visual and other artistic works. The increased attenuation of human creativity from the actual generation of the final work will prompt challenging questions regarding how much human input is necessary to qualify the user of an AI system as an ‘author’ of a generated work.”[xiii] Also, Judge Beryl Howell cited cases where copyright protection was denied for celestial beings, a cultivated garden, and a monkey who took a selfie.[xiv] Therefore, the Court upheld the USCO’s decision to deny Thaler’s copyright registration, stating that AI-generated artwork lacks “human authorship,” and cannot be the subject of a valid copyright claim. The opinion stressed, “Human authorship is a bedrock requirement…Human involvement in, and ultimate creative control over, the work at issue was key to the conclusion that the new type of work fell within the bounds of copyright.”[xv]
Also, Judge Beryl Howell explored the purpose of copyright law, which is to encourage humans to engage in creation. Copyrights and patents were conceived as “forms of property that the government was established to protect, and it was understood that recognizing exclusive rights in that property would further the public good by incentivizing individuals to create and invent…The act of human creation—and how to best encourage human individuals to engage in that creation, and thereby promote science and the useful arts—was thus central to American copyright from its very inception.”[xvi] In other words, copyright law was not designed to reach nonhuman actors which “need no incentivization.”[xvii]
Copyright law is preemptive and exclusively governs the subject matter of claims that are within the scope of the Copyright Act. Since this case does fully address AI-generated works with no human involvement, advertising companies and the entertainment industry should grow keenly aware that it is highly possible that entirely AI-generated advertisements and creative works may not be subject to copyright protection. Without property rights to infringe, there may be no legal basis to prevent third parties from using creative works completely generated by AI. This could potentially create legal problems on ownership and discourage wide applications of AI-generated works in the creative industry. Also, not allowing AI-generated inventions to be patented would incentivize AI owners to rely on gamesmanship with patent offices by failing to declare a filing is based on AI-generated invention. In contrast, allowing patent rights and intellectual property protection for AI-generated output would encourage the disclosure of AI involvement in inventions and provide more accurate information to future inventors.
It is important to note that this decision does not stretch to whether a human could be considered as a co-author of AI-generated works. In other words, the court did not directly address the question of how much human input is necessary to qualify the user of an AI system as the “author” of a generated work.[xviii] Judge Beryl Howell suggested AI’s expanding role in generating works would necessitate revisiting copyright law’s human requirement. In the middle ground between complete human authorship and complete AI-generated content lies an important and intriguing legal question: how much human input into an AI program would be reasonable and fair to constitute human authorship? How to assess the originality of AI-generated content that comes from systems trained on existing copyrighted works? As AI-generated works become more sophisticated and prevalent in the future, either the courts or Congress will have to address the complicated questions regarding AI’s involvement in creative works, which could potentially redefine and renegotiate the future legal landscape of copyright protection, intellectual property rights, and ownership.
In March of 2023, the USCO issued a statement of policy on AI indicating that works including “human-authored elements combined with AI-generated images” are copyrightable.[xix] On September 27, 2023, Shira Perlmutter, Director of the USCO, revealed the Office’s intentions in a House Senate Subcommittee hearing as “to guide people as to how they can register works that include generative AI outputs. They should disclaim the portion of the work that was generated by AI, and we will issue a registration for the work as a whole, where there is some human authorship.”[xx]
On June 7th, 2023, Stephen Thaler’s attorney, Professor Ryan Abbott shared his thoughts on the future of AI and IP concerns at a Senate Subcommittee hearing.[xxi] In his testimony, Professor Ryan Abbott talked about the challenges faced by the Artificial Inventor Project in filing patents. He asked Congress to consider modifying the Patent Act to allow the protection of AI-generated inventions by prohibiting patent ability to be determined based on how an invention is made.[xxii]
Australian Court Finds AI System Can Be “Inventor”
Simultaneously, Stephen Thaler has applied for patents listing his AI called DABUS (“device for the autonomous bootstrapping of unified sentience”) as sole inventor in other countries since 2018 and received different outcomes.[xxiii] The UK Intellectual Property Office (UKIPO) and European Patent Office (EPO) rejected the applications because an inventor on a patent application must be a “natural person” who have “legal capacity,” meaning the “ability, according to a source of law, to be the subject of rights and duties.”[xxiv] In contrast, Australia and South Africa put less emphasis on the necessity of human authorship for copyright and patent protection. In 2021, Justice Jonathan Beach of the Federal Court of Australia found that artificial intelligence could be considered as an “inventor” for the purposes of the Australian patent regime and the plain language of the Australian Patents Act contained no prohibition on AI inventorship.[xxv] Under Section 15 of the Patents Act, a patent for an invention may only be granted to a person who:
- is the inventor; or
- would, on the grant of a patent for the invention, be entitled to have the patent assigned to the person; or
- derives title to the invention from the inventor or a person mentioned in paragraph (b); or
- is the legal representative of a deceased person mentioned in paragraph (a), (b) or (c).
Justice Beach found that there was “no specific provision [in the Patents Act] that expressly refutes the proposition that an artificial intelligence system can be an inventor.”[xxvi] While the Commissioner of Patents sought to emphasize the dictionary definitions of “inventor” (given that “inventor” is not defined in the Act), Justice Beach ruled that there is no basis to exclude AI from being considered as an “inventor” within the scope of the Patents Act, or to “preclude a class of otherwise patentable inventions from patentability on the basis of an exclusion that is not apparent from the express words of the Act. Indeed, that would be the antithesis of promoting innovation.”[xxvii]
AI Gets Its First Patent in South Africa
On June 24, 2021, South Africa’s patent office granted DABUS, an AI system created by Stephen Thaler, which stimulates human brainstorming and creates new inventions, two patents. It reasoned that human intervention in AI programming and training fulfilled the inventorship requirement.[xxviii] The notice of issuance was published in the July 2021 Patent Journal.[xxix] However, South Africa’s High Court overturned this decision, finding AI insufficiently autonomous to qualify under the Patents Act without human inventorship or assignment. Stephen Thaler and his team appealed to the Constitutional Court arguing an invention’s patentability should not solely depend on its creator’s identity.[xxx] Historically, an inventor of a patent has to be a human, although the ownership of that patent is commonly granted to the company that employs the inventor. While patent law in most jurisdictions is very specific in how it defines “inventor”, Stephen Thaler and his team argue that the status quo does not fit the purpose of the Fourth Industrial Revolution.[xxxi] The case remains pending.
Conclusion
In 1925, Judge Benjamin Cardozo spoke at a graduating law school class that “the new generations bring with them their new problems which call for new rules, to be patterned, indeed, after the rules of the past, and yet adapted to the needs and justice of another day and hour.”[xxxii] This is the case for AI. Advances in AI are testing the boundaries of established IP law. AI systems can now generate creative outputs including art, music, and inventions without direct human involvement. This challenges the assumption in copyright and patent law that protected works require human authorship.
Today, it is not entirely clear whether AI-generated inventions, those made without traditional human inventors, are eligible for intellectual property protection. At least for now, IP law remains firmly based in the assumption of human creators despite technological advancement. The tension and dynamics between intellectual property law doctrine that requires human authorship and the reality of AI’s increasing capabilities to generate creative works and inventions autonomously will become more relevant and sophisticated in the future. Resolutions wrestle with conceptual puzzles at the intersection of personhood theory and policy consideration to promote innovation. As the ability of AI to invent increases exponentially, it might surpass humans whose abilities to increase inventive capabilities are far more limited.[xxxiii] Listing an AI as an inventor is not a matter of providing rights to machines, but it would protect the rights of traditional human inventors and the integrity of the intellectual property system.[xxxiv] The proliferation of AI will likely compel renegotiating and reassessing how IP law should adapt to emerging technologies and societal needs, as AI is a significant issue in terms of investment and protection in art and technology. Based on the ongoing litigation in various jurisdictions, it seems possible that AI-generated materials might be eligible for copyright in the future.
Image Source: susan-lu4esm, https://pixabay.com/illustrations/ai-generated-robot-technology-8015425/
[i] U.S. Const. art. I § 8, cl. 8.
[ii] 17 U.S.C. § 102(a).
[iii] 17 U.S.C. § 101.
[iv] https://www.wipo.int/wipo_magazine/en/2019/06/article_0002.html
[v] Id.
[vi] https://artificialinventor.com
[vii]Stephen Thaler v. Shira Perlmutter and The United States Copyright Office, D.D.C., No. 1:22-cv-01564, P.3. (June 2, 2022).
[viii] Id.
[ix] Id.
[x] Id.
[xi] Id. at 5.
[xii] Id. at 8.
[xiii] Id. at 18.
[xiv] Id. at 16.
[xv] Id. at 11.
[xvi] Id. at 13.
[xvii] Id.
[xviii] District Court Rules that AI-Generated Works Cannot Be Copyrighted. Morrison Foerster Client Alert. (29 Aug 2023). See: https://www.mofo.com/resources/insights/230829-district-court-rules-that-ai-generated-works
[xix] Martin Gomez & Ashleigh Nickerson. Is It Possible to Copyright Works That Include AI-Generated Material? The Answer Seems to Be Yes—But Only When “Authorship” of the Work Can Be Attributed to A Human. Goodwin Insight. (October 2, 2023).
[xx] Oversight of the U.S. Copyright Office. Subcommittee on Courts, Intellectual Property, and the Internet. (September 27, 2023).
[xxi]https://www.senate.gov/isvp/?auto_play=false&comm=judiciary&filename=judiciary060723p&poster=https://www.judiciary.senate.gov/assets/images/video-poster.png&stt=0
[xxii] Id.
[xxiii] Victoria Basham. Australian Court Agrees with South Africa that AI Can Be Inventors. The Global Legal Post. (July 30, 2021).
[xxiv] Meshandren Naidoo. In a World First, South Africa Grants a Patent to an Artificial Intelligence System. Quartz. August 9, 2021.
[xxv] Thaler v. Commissioner of Patents. FCA 879. 2021.
[xxvi] Id.
[xxvii] Id.
[xxviii] Meshandren Naidoo. In a World First, South Africa Grants a Patent to an Artificial Intelligence System. Quartz. August 9, 2021.
[xxix] Companies and Intellectual Property Commission. CIPC Intellectual Property Online: https://iponline.cipc.co.za/Publications/JournalPublications.aspx
[xxx] Miguel Bibe. DABUS: the “Natural Person” Problem. World Intellectual Property Review. September 17, 2021.
[xxxi] World’s First Patent Awarded for an Invention Made by an AI Could Have Seismic Implications on IP Law. The University of Surrey Press Release. July 28, 2021.
[xxxii] Ryan Abbott. The Reasonable Robot: Artificial Intelligence and the Law. Cambridge University Press. 2020. P.3.
[xxxiii] Id. at 12.
[xxxiv] https://www.wipo.int/wipo_magazine/en/2019/06/article_0002.html