Now You See Me is a Hollywood trilogy built on spectacle, that is, magicians who blur the lines between reality and illusion, leaving audiences questioning what they just witnessed. The entire premise rests on deception that feels real, even when we know it isn’t. Artificial Intelligence (“AI”) operates in a strikingly similar way.
The illusion AI creates is not of disappearing coins or impossible escapes, but of something far more profound: human-like intelligence. It writes, paints, composes, and even decides often so convincingly that it appears almost sentient. But is it?
Sentience, in its simplest sense, refers to the capacity to feel, perceive, and experience. AI, despite its sophistication, does not truly possess this human quality. And yet, its outputs simulate it so effectively that it becomes difficult to distinguish between genuine creativity and algorithmic imitation.
This is where AI exists, in a liminal space between sentience and illusion. What emerges from this dual characteristic of AI is a powerful illusion of creativity, one that challenges our traditional assumptions about authorship and originality of content. It is within this blurred boundary between reality and illusion that the story of Dr. Stephen Thaler and his AI system, DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), unfolds.
DABUS is an AI system that is designed to create autonomously, algorithmically producing inventions and artworks without any direct human input. One such creation is the artwork called A Recent Entrance to Paradise, which lies at the heart of a legal controversy questioning the limits of copyright law.
When Dr. Thaler sought copyright protection for this autonomous artwork in the United States, the answer was unequivocal. The courts held that authorship under the U.S copyright law is inherently human, and works generated autonomously by machines fall outside its protection.
Undeterred, Dr. Thaler has now brought the same question to India, and this question tests the boundaries of the Copyright Act, 1957 (“Act”) at a moment when AI is beginning to strain legal frameworks built fundamentally around natural persons.
The Delhi High Court has directed the Copyright Office to adjudicate Dr. Thaler’s application. It is Dr. Thaler’s suggestion that the author of the artwork is DABUS and not him. At one level, the response to this suggestion can be straightforward: adopt the approach of the U.S. courts and reject the claim at the threshold. After all, it is a fairly settled position in Indian law that authorship must vest in a natural person. This is not only borne out by judicial precedent but is also fortified by the plain language of the Act. Section 2(d)(vi), for instance, defines the author of a computer-generated work as the “person who causes the work to be created.” The use of the term person necessarily excludes an AI system or algorithm from being recognised as an author.
However, now that the Copyright Office has been tasked with examining Dr. Thaler’s application, the inquiry must go beyond questions of authorship. It must entail dealing with a more fundamental question of whether fully autonomous AI-generated works can meet the requirement of originality under the Act.
Under Section 13 of the Act, copyright subsists only in original artistic works. Most AI systems function through iterative processes which involve analysing vast datasets, identifying patterns, and generating outputs based on those patterns. While the results may appear novel, the underlying process is arguably one of sophisticated reproduction rather than genuine creation. The absence of conscious human engagement raises doubts as to whether such outputs embody the modicum of creativity required under Indian copyright law.
Take, for instance, a scenario where DABUS generates an artwork by collating elements from works of art available in the public domain. Even if the final output appears unique, it may still fall short of the originality threshold. This is because it is well-settled that mere selection and arrangement of pre-existing material particularly where it is routine or lacks creative judgment does not qualify as an original work capable of protection under the Act.
Therefore, under the current framework of the Act, copyright protection for fully autonomous AI-generated works is unlikely to be recognised. The emphasis on human authorship and originality grounded in a modicum of creativity creates a gap that such works cannot bridge. However, while Mr. Thaler’s claim may fail, it raises deeper questions about the limits of legal frameworks designed for natural persons, now being tested by the growing sophistication of AI systems. Under the Act these questions include who qualifies as the author where human involvement is minimal, whether originality can subsist in iterative outputs, how liability should be assigned for infringing AI-generated works, and whether new categories of rights are needed for machine-generated creations.
In this sense, the challenge posed by AI is not merely about recognising a new kind of author, or recalibrating creativity but more fundamentally, rethinking the extent of statutory reform required in a world increasingly shaped by the illusion of sentience that AI so convincingly creates.
Nakul Dewan is a Senior Advocate and King’s Counsel.