Author: Karuna Soni, K.G shah Law School, SNDT University
Abstract
In this study, the legal conflict arising from the tension between the static tenets of intellectual property laws and the dynamic nature of the generative AI technology is evaluated. Through the review of seminal court decisions in the US, Europe, and India, the history of the doctrine of the “human author” is explored. The unclear boundaries of “substantial human involvement” needed to make valid copyright claims using AI technology are delineated, along with the risks associated with the lack of protection of AI-generated material in the public domain.
To the Point
However, the rapid rise of generative Artificial Intelligence (AI) technology has triggered an existential crisis in intellectual property law frameworks. The central issue that defines the legal fight is simply: Is it possible for a non-human to have copyright and does it make sense for autonomous machine output to qualify as legally original?
At present, most of global legal doctrine assumes that copyright must always be inherently linked to human authorship. The work created by autonomous AI technology is classified either as uncopyrighted works that are automatically part of the public domain, or else a mere adaptation of digital work without any protection under copyright. This generates a grave economic dilemma as huge investments are being made in algorithmic automation, but the underlying economic value—the product itself—is still indefensible.
Use of the Legal Jargon
Sui Generis: Latin for “of its own kind.” It stands for a legal framework entirely novel for addressing a unique problem. Legal scholars believe that rather than twisting centuries-old copyright frameworks to include algorithms, lawmakers should craft a sui generis right solely for AI data productions.
Modus Operandi: Latin for “method of working.” In IP lawsuits, examination of the modus operandi of the AI system is necessary for the court to determine whether the device simply serves as an auxiliary (for instance, the camera) or is indeed the determinant force behind the choices made expressively.
De Minimis: Latin for “of minimal things.” The law is indifferent to small details. With respect to generative content, inputting a simple textual prompt (“draw a blue cat”) is regarded as de minimis human contribution that falls short of meeting the bar of creativity needed for copyright protection.
Sine Qua Non: Meaning “without which, not,” in Latin. An essential requirement. Intellectual effort and conscious will on the part of humans are the sine qua non of intellectual property protection in the entire world.
Locus Standi: Meaning “place of standing” in Latin. Right of a party to have their case heard in court. As artificial intelligence does not have a legal personality, it lacks locus standi to sue for copyright infringement or have property rights.
Animus Domini: Meaning “intention to own” in Latin. Refers to a conscious intention in the mind regarding ownership of a piece of property. As algorithms lack sentient consciousness, they do not have animus domini.
The proof
The denial of copyrights for non-human systems is not a whimsical response to current technological developments; it is simply an application of the “Doctrine of Originality” and “Expression-Idea Dichotomy.” Copyright protects the expression of an idea but not the idea itself.
When a human provides a creative text prompt to a large language model or a diffusion network, he is merely giving an unprotected idea. This is because an idea is any abstract concept of a theme or style. Then the machine converts the idea into tangible form by performing calculations of the pixel weight, token hierarchy, and structural semantics according to its neural network training.
Since the machine performs the creative choices and not the human—such as syntax, layout, color, or melodies—the link between the creative choices of the human brain and the tangible end product is eliminated. As a result, the product does not show any traces of human creativity, which prevents it from being protected under global copyright laws.
Case Laws
1. Burrow-Giles Lithographic Co. v. Sarony (1884) – United States
The Background: The defendant lithographic company reproduced the portrait of author Oscar Wilde made by the plaintiff photographer Napoleon Sarony. In its defense, the lithographer claimed that a camera is an instrument which automatically records the natural light rays and cannot be treated as a product of intellectual conception by humans.
The Legal Decision: The U.S. Supreme Court disagreed, setting a historical precedent for copyrighting photos. The Supreme Court decided that photos were subject to copyright protection since they reflected the original conceptions of human authors. Sarony himself directed the lighting, picked up the exact costume and pose for the picture. This case established the international rule that the use of a mechanical means did not deprive the work of copyrightability when the role of the human mind was the dominating one.
2. Naruto v. Slater (2018) – “Monkey Selfie” Case
The Background: A rare species of crested macaque named Naruto encountered a camera which had been left unattended by the British photographer David Slater. Having shown interest in the camera’s lens, Naruto pressed the shutter button and created a self-portrait of unprecedented clarity and brilliance. After that, PETA filed a suit against Slater, arguing that the copyright of the photo belongs to the monkey, who should be paid all royalties from its commercial use.
The Legal Decision: The US Ninth Circuit Court of Appeals concluded that although the photo is unique and original in every way possible, animals, as well as any other entities apart from humans, do not have any standing according to the Copyright Act. It is important to note that statutory language such as “children,” “widows,” and “heirs” clearly implies that Congress intended to create a legal framework for copyrighting solely for human individuals.
3. Thaler v. Perlmutter (2023) – United States
The Background: Dr. Stephen Thaler created a piece of two-dimensional artwork named “A Recent Entrance to Paradise” using his own neural network, which he named the “Creativity Machine.” On purpose, he registered the copyright application under the name of the AI as the main author and himself as the owner of the work through the “work-for-hire” doctrine because, otherwise, the denial of the copyright would defeat the economic aim of IP law.
The Legal Decision: The decision rendered by the District Court of the United States of America for the District of Columbia upheld the rejection by the United States Copyright Office of the copyright application. Judge Beryl Howell noted that human authorship was one of the cornerstones of copyright. It has been established that copyright was specifically designed as an economic incentive for motivating people in creating things; an algorithm does not feel any economic incentives, does not have financial needs and cannot be motivated by statutory monopoly.
4. Ankit Sahni vs. Indian Copyright Office (2020 – Present) – India
Background: Indian artist and intellectual property lawyer Ankit Sahni made use of an AI-powered paint tool referred to as “RAGHAV” (Robust Artificially Intelligent Graphics and Art Visualizer) for producing an art piece entitled “SURYAST”. The creation of “SURYAST” involved providing a photograph of his own creation into the software, adding Vincent Van Gogh’s “The Starry Night” as a style input, and controlling the intensity of style transfer through a sliding scale.
The Legal Judgment: As an unprecedented action on a global scale in November 2020, the Indian Copyright Office granted Sahni’s request for registration of “SURYAST”, co-authoring the piece with the AI tool “RAGHAV”. However, soon realizing the extensive effects this decision would have on the system, in late 2021 the Copyright Office withdrew the application, casting doubts on the legitimacy of the AI tool, and pointing out that according to Section 2(d) of the Indian Copyright Act, 1957, only a person can be recognized as an author. Although the final legal outcome is still uncertain due to ongoing procedure back-and-forth, this case fully illustrates the huge regulatory gap in co-authorship channels.
5. Infopaq International A/S v. Danske Dagblades Forening (2009) – European Union
Background: A case dealing with automated data clipping and text-scraping algorithms used in obtaining 11-word snippets of text from newspapers. It was necessary for the court to decide on what minimum quantity of information qualifies as an “original work.”
Judgment: The European Court of Justice ruled that only an item is eligible for copyright protection if it is original in the sense of being “the author’s own intellectual creation.” This implies that such a state can be attained only when the creator makes free, conscious, and creative decisions, thereby putting the definite imprint of “his personality” on the finished work. Such an approach could not be employed by automated data processing systems using mathematical probability matrices.
Conclusion
The convergence between the generative AI technology and conventional intellectual property laws reveals significant structural discord. The statutory provisions that were created in industrial times to safeguard human efforts from mechanical duplication fall apart when the duplication technology becomes the creator.
If the world courts insist on their absolutism in the insistence of human authorship, they may end up discouraging commercial organizations from venturing in any extensive automation projects. On the other hand, embracing the unrestricted copyrighting by machines means that large tech companies can create millions of copyrightable material within an hour and shut off the public domain forever.
It is necessary to shift our focus from the binary approach to forcing the AI into our human statutes and for legislatures around the globe to work ahead and formulate an appropriate Sui Generis IP regime – possibly a short term protection for the human investor or controller of the AI system – not for the piece of computer programming.
FAQ
Q1: Can I copyright my artwork, essay, or line of code even when created using the services of ChatGPT or Midjourney?
Ans: As per the existing international laws on copyrighting, you do not have copyright protection rights on the outputs generated through an AI-based autonomous system in response to your textual prompt. However, if you take the output of such an AI program as a base and make substantial human modifications to the output in the form of digital painting, rewriting, or incorporation in a layout driven by humans, then you might copyright your modifications but not the outputs of the AI.
Q2: Is the rejection of AI output by the copyright office sufficient ground for others to freely steal and reproduce it?
Ans: Yes. If the creation is found to be an AI product with little human intervention involved, then it fails to have any human creator. In such a situation, it fails to have any proprietary rights assigned to it. This places it in the public domain and gives all other competitors the legal right to download, modify, copy, and distribute the work.
Q3: What qualifies as “sufficient human intervention” to make an AI assisted project copyrightable?
Ans: There is no clear cut off or mathematically quantifiable percentage, but the key is who ultimately makes the expressive decision. Choosing a particular style, making a basic settings adjustment, and entering a complex text prompt into an AI would not qualify as sufficient human intervention because the machine decides on the final pixel by pixel output. On the other hand, using an original photograph of your own creation, running it through an AI model with a unique texture filter, and manually changing its colors and arrangement in the editing software would qualify.
Q4: Is it legal for AI companies to train their LLMs and image generators on copyrighted data without permission?
Ans: It is currently the subject of several billion-dollar lawsuits around the world. AI companies claim that data scraping falls under “Fair Dealing” or “Fair Use”, since the machine doesn’t just copy and paste the artwork, but simply recognizes abstract patterns mathematically to learn how different concepts appear or sound. Content creators argue that it effectively devalues their commercial product. Final, definitive high-court rulings on this systemic issue are still evolving globally.
