“What is originality?
Undetected Plagiarism.”[1]
– William Ralph Inge
The idea of originality has always been confusing. Jonathan Lethem once pointed out that when people call something original, nine times out of ten, they simply don’t know the source it drew from. He argues that Nothing truly comes from nowhere.
In the field of creativity, particularly in fashion, the distinction between inspiration and imitation has never been well defined. Designers often borrow from existing trends, cultural references, and past creations, then rework with alterations, which blurs the distinction between what is truly “original” and what is merely reinterpreted. While such ambiguity may be acceptable in artistic discourse, the law does not afford the same flexibility. In the upcoming paragraphs, you will know about the copyright dilemma in AI fashion.
The famous phrase attributed to Pablo Picasso-“good artists copy, great artists steal” may sound convincing in creative discourse, but it oversimplifies a far more complicated reality. While “stealing” is often referred to as a metaphor for inspiration, philosophically, it may appear harmless, but it is far less straightforward in law. In copyright jurisprudence, concepts like originality and copying are neither fluid nor philosophical notions, but legally, they aren’t vague ideas, carrying serious weight, one that determines real consequences for authorship, ownership, and protection.
Traditionally, copyright law has been closely associated with originality in human creativity, demanding independent skill, judgment, and intellectual effort.[2] However, the boom of artificial intelligence in recent years, particularly in fashion, has begun to disrupt its very core, which is not easy to ignore. Because artificial intelligence does not just assist anymore, with just a few clicks, it can generate, alter, and replicate designs with minimal or no human involvement[3], thereby complicating the legal understanding of authorship itself. What once required years of artistic labour can now be mimicked within seconds through prompts. More critically, these systems are trained on vast datasets of existing creative works, raising concerns not only of authorship but also of indirect appropriation and unacknowledged borrowing.
This raises a fundamental legal dilemma: can a design created without human intervention satisfy the requirement of originality? If not, who owns it? and whether these AI-assisted imitations actually constitute infringement or are just another loophole for bypassing the law?
There is a tendency to confuse originality with novelty. In copyright jurisprudence, originality is not concerned with novelty or uniqueness in the absolute sense, but the quality or state of being the product of independent creation and having a minimum degree of creativity.[4] The degree to which a product claimed for copyright is the result of an author’s independent efforts[5]
Going by this definition, originality as a requirement for copyright protection possesses a lesser standard than novelty[6] under patent law[7], such that to be original, a work does not have to be novel or unique.[8]Copyright law[9] tolerates similarity so long as the work is not directly copied. In other words, two works may appear similar, yet both may qualify for protection if they are independently created. The law, therefore, does not prohibit inspiration, but it draws a line at replication.
This principle evolved through judicial scrutiny. Earlier approaches, particularly the “sweat of the brow” doctrine as seen in Walter v. Lane[10], emphasized the labour and effort invested in a work. This approach was later reconsidered for being overly protective of effort rather than creativity.
A sharper, refined standard emerged in Feist Publications Inc. v. Rural Telephone Service Co.[11], where the Court held that originality requires independent creation coupled with a minimal degree of creativity, thereby rejecting the idea that mere labour is sufficient. This significant shift emphasizes not effort alone, but creative contribution. This distinction becomes particularly significant in the context of artificial intelligence, where outputs may satisfy the appearance of creativity, yet lack the underlying human intellectual effort required by law.
Later, the “skill and judgment” test from CCH Canadian Ltd. v. Law Society of Upper Canada[12] brought balance. It clarified that originality requires intellectual effort that is neither purely mechanical nor trivial. In Bleistein v. Donaldson Lithographing Co.[13], The Court cautioned against judicial evaluation of artistic merit, holding that even simple works may qualify for protection, raising further questions about how such standards apply to AI-generated outputs.
In the Indian context, the Supreme Court in Eastern Book Company v. D.B. Modak[14] adopted this balanced approach, rejecting the “sweat of the brow” doctrine and affirming that originality requires a minimum level of creativity involving skill and judgment. Applying this standard to AI-generated works, outputs produced with minimal human involvement may fail to meet the requirement of skill and judgment, thereby falling outside the scope of protection.
Each doctrine differs in how much effort, creativity, or intellectual input is required for a work to be considered original. The point of incidence of the originality is the owner. Originality is not always about the origin of the idea but about the way in which it is presented or expressed. Section 13[15] outlines the types of works that are eligible for copyright protection. The significance of originality is highlighted in the case of Aggrawal Publishing House vs Board of High School Intermediate Education[16]
A look that appears on a runway or Instagram can be scanned, altered slightly, and reproduced halfway across the world in hours. For a designer who has spent years developing a garment, protecting it with trademarks, copyright in sketches, and registration under the Designs Act[17], the risk is huge. Once it’s public, someone else can feed it into AI, change a few elements or proportions, and create something that looks different enough to dodge direct infringement while still capturing the original’s commercial feel.
This immediately raises a tough question: who counts as the author? The person giving prompts? The developer who built the system? Or the AI itself? Existing frameworks give no clear answer. In Naruto v. Slater[18], the court made it plain that authorship is inherently human. Under Indian law, authorship generally belongs to whoever exercises creative control. But with AI, that control often gets diluted because the final result is highly uncertain and depends heavily on computational processes.
Yet here’s the paradox: AI-generated works may not get protection themselves, but they can still trigger infringement liability if they resemble protected designs too closely. AI can’t own rights, but it can help violate them. This creates a real imbalance. Original designers face growing difficulty stopping AI-enabled imitation, especially when modifications fall short of the “substantial similarity test”.[19]Meanwhile, people using AI with little personal effort can still profit.
In India, the problem gets even messier because of our price-sensitive market. Most consumers want affordable, trending clothes rather than exclusive originals. Cheap knock-offs and “inspired” dupes flood local markets. AI makes imitation faster and subtler, turning what was once manual copying into scaled, sophisticated derivatives that sit just inside legal boundaries. Enforcement is already weak, small manufacturers are hard to track, litigation is expensive, and many smaller creators can’t afford to fight.
In the end, AI in fashion forces us to face a basic question: can there truly be design without a real designer in the eyes of the law? Right now, the framework recognises authors but struggles with algorithms, protects effort but not automation, and values originality but finds it harder to defend when imitation becomes so easy. Without clearer rules on human involvement, better infringement tests for algorithmic changes, and more transparency around AI training, the gap between legal theory and ground reality will only keep widening, especially in India’s fast-fashion-driven market.
Author: Aditya Gautam, Final Year Student at the Faculty Of Law, University Of Lucknow, India
Read Similar Article: Apple Watch Ultra 3 – Patents, Trademarks & Design IP
1. Austin Kleon, Steal Like an Artist: 10 Things Nobody Told You About Being Creative, pg. 7 (Workman Publ’g Co. 2012) (attributing the quote to William Ralph Inge)
2. E. Book Co. v. D.B. Modak, (2008) 1 S.C.C. 1 (India).
3. Patrick Zurth, Artificial Creativity? A case against Copyright Protection for AI-Generated Works, 25 UCLA Journal of Law and Technology(Spring 2021).
4. Black’s Law Dictionary (8th ed. abr. 2005)
5. Bryan A. Garner, Black’s Law Dictionary, p. 929. (Abridged 8th edn, St. Paul: Thomsom/West, 2005) See also Bryan A. Garner, Black’s Law Dictionary 1327 (11th ed. 2019)
6. The Patents Act, 1970, § 2(1)(l), No. 39, Acts of Parliament, 1970 (India)
7. The Patents Act, 1970, No. 39, Acts of Parliament, 1970 (India).
8. Ibid.
9. The Copyright Act, 1957, No. 14, Acts of Parliament, 1957 (India)
10. Walter V. Lane, A.C. 539 (HL).
11. Feist Publications Inc. v. Rural Telephone Service Co., 499 U.S. 340(1991)
12. CCH Canadian Ltd. v. Law Society of Upper Canada, 1 SCR 339, 2004 SCC 13
13. Bridgeman Art Library v. Corel Corporation,36 F. Supp. 2d 191(S.S.N.Y. 1999)
14. Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1.
15. The Copyright Act, 1957, § 13, No. 14, Acts of Parliament, 1957 (India).
16. Aggrawal Publishing House v. Board of High School Intermediate Education, AIR 1971 SC 155.
17. The Designs Act, 2000, No. 16, Acts of Parliament, 2000 (India).
18. Naruto v. Slater, 888 F.3d 418 (9th Cir. 2018)
19. Substantial similarity, including the subjective question of whether there is a shared total concept and feel, can be decided by a court