202611.03
37

AI Can Create, But Cannot Author: India’s Copyright Office Rejects DABUS Authorship

Indian copyright law may protect certain AI-generated outputs, but it does not presently recognise AI systems as authors. For computer-generated works, authorship remains tied to the legally recognised person who causes the work to be created.

The growing use of generative artificial intelligence has raised fundamental questions of authorship, ownership and copyright protection. In a landmark decision concerning AI-generated works, the Registrar of Copyrights, Prof. (Dr.) Unnat P. Pandit, rejected Dr. Stephen L. Thaler’s application to register copyright in the artwork “A Recent Entrance to Paradise”, where the AI system DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) was named as author. The Copyright Office held that the artwork could, in principle, satisfy the originality requirement under the Copyright Act, 1957, but that DABUS could not be recognised as an author because it is neither a natural person nor a juristic person recognised by law.

The order is significant because it separates two questions often conflated in AI copyright debates a) whether a machine-generated output can be original, and b) whether the machine itself can be named as author. The first was answered favourably in principle; the second was answered in the negative. On the facts, that person was Dr. Thaler, not DABUS; however, because the application continued to name DABUS as author, it was rejected.

Background of the Case:

  1. Dr. Thaler sought copyright registration for the AI-generated artistic work “A Recent Entrance to Paradise”, naming DABUS, an AI system developed by him, as the author.
  2. Dr. Thaler claimed ownership based on his development and control of DABUS, while maintaining that DABUS autonomously generated the final artwork. The Copyright Office raised objections on originality, authorship and ownership, conducted hearings, and appointed an Amicus Curiae to assist on these novel issues.

Applicant’s Contentions:

  1. Autonomous Generation by DABUS: The Applicant argued that DABUS generated the artwork independently through a structured neural-network process, while Dr. Thaler’s role was limited to designing, configuring and training the system.
  2. Section 2(d)(vi) Covers Computer-Generated Works: The Applicant relied on Section 2(d)(vi), under which the author of a computer-generated work is “the person who causes the work to be created”, arguing that this provision should accommodate autonomous AI outputs.

Amicus Curiae’s Submissions:

The Amicus Curiae submitted that AI-generated works may qualify for copyright protection, but authorship must vest in a legally recognised person. It emphasised that originality and authorship are separate issues: a work may be original even if the AI system that generated it cannot be the legal author.

Authorship Vests in the Person Causing Creation: The Amicus argued that Section 2(d)(vi) attributes authorship to the person who conceived, configured, deployed or otherwise caused the generative process to occur. Relying on the “mastermind” or “controlling intelligence” test, the focus should remain on the person exercising creative control, not the technological instrument used to execute the work.

Copyright Office’s Findings:

  1. The Work Satisfied the Originality Requirement: The Copyright Office held that originality under Section 13 does not require novelty or inventiveness. Relying on Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1, the Registrar concluded that the artwork satisfied the minimum degree of creativity required for copyright protection.
  2. The Registrar observed that the artwork had not been shown to be copied from any pre-existing work and possessed sufficient expressive character. The use of AI technology did not, by itself, make the work unoriginal.
  3. DABUS Cannot Be Recognised as an Author: The central issue before the Copyright Office was whether DABUS could be recognised as the author of the work.

    The Registrar held that Section 2(d)(vi) attributes authorship to “the person who causes the work to be created”, not to the machine that generates the output. The provision therefore focuses on legally attributable causation rather than the immediate mechanism of production.

    On that basis, DABUS was treated as a technological tool and not as the legal author. It could not qualify as the person who had “caused the work to be created” within the meaning of the Act.

  4. Dr. Thaler was the “Person who caused the work to be Created: The Registrar observed that Dr. Thaler conceived and configured DABUS, supplied the visual and linguistic inputs, established the operational parameters and initiated the process through which the artwork came into existence. These actions established a sufficiently direct nexus between Dr. Thaler and the creation of the work.

    The Registrar concluded that Dr. Thaler’s conception, configuration, inputs, parameters and initiation of the process established a direct nexus with the creation of the work. Dr. Thaler, and not DABUS, was therefore the person who caused the work to be created.

  5. Ownership Claim Was Legally Inconsistent: The Registrar further held that the application suffered from a defect because it identified DABUS as author while simultaneously claiming Dr. Thaler as owner.

    The Copyright Office also found the ownership claim inconsistent: DABUS was named as author while Dr. Thaler claimed ownership. Since DABUS is not capable of owning, assigning or transferring copyright, the proposed separation between AI authorship and human ownership was legally unsustainable.

  6. No Recognition of DABUS as a Technological Generator: As an alternative, the Applicant requested that Dr. Thaler be recognised as author while DABUS be identified in the Register as the technological generator of the work. The Registrar declined to decide that issue in the abstract, noting that no amended application had been filed, the request was conditional, and no separate, duly formulated request for a legally neutral remarks entry had been placed before the Office.

Key Takeaways:

  1. AI-generated works are not automatically excluded from copyright protection under Indian law: The Copyright Office accepted that a work generated through AI can satisfy the originality requirement under the Copyright Act, 1957.
  2. Originality and authorship are distinct legal enquiries. The decision clarifies while an AI-generated work may qualify for copyright protection, the question of who can be recognised as the author must be determined separately.
  3. Section 2(d)(vi) attributes authorship to the legally recognised person who causes the work to be created: The Copyright Office held that authorship vests in the legally recognised person responsible for causing the work to come into existence, rather than in the AI system generating the output.
  4. AI systems cannot presently be recognised as authors under Indian copyright law considering AI system such as DABUS is neither a natural nor a juristic person capable of being recognised as an author under the Copyright Act.
  5. Human involvement remains central to copyright ownership and attribution: Where AI is used as a creative tool, copyright law continues to look to the human who conceived, configured or otherwise caused the work to be created.
  6. Any recognition of AI authorship would require legislative intervention: The question of granting legal personality or authorship status to AI systems is a matter of legislative policy and would require statutory amendment rather than administrative or interpretative expansion of Section 2(d)(vi).

Position in the United States:

Dr. Thaler’s corresponding application in the United States was refused because the work was presented as having been created autonomously by the “Creativity Machine,” with the machine named as the sole author. The U.S. Copyright Office, and later the District Court and the D.C. Circuit, held that copyright protection under the U.S. Copyright Act requires human authorship in the first instance. Since the application identified a non-human machine as author, and Dr. Thaler’s alternative claim that he should be treated as author by virtue of creating or using the machine had not been properly preserved before the agency, the refusal was affirmed.

Comparison with the Indian Position:

The Indian and U.S. decisions arrive at the same core conclusion that an AI system cannot presently be recognised as an author. The difference lies in the route taken. The U.S. approach treats human authorship as a threshold requirement for copyright protection, whereas the Indian Copyright Office distinguished between originality and authorship, accepting that an AI-generated work may, in principle, be original, but holding that authorship must vest in the legally recognised person who caused the work to be created. This makes the Indian position somewhat more accommodating of AI-generated outputs, while still refusing legal authorship to the AI system itself.

Our Comment:

Viewed against the U.S. approach, the Indian order is significant because it refuses AI authorship without shutting the door on copyright protection for AI-generated works. The Copyright Office has drawn a principled distinction between copyrightability and authorship. An AI-generated output may be original, but authorship must still vest in a legally recognised person.
The approach is pragmatic. It allows the existing Copyright Act to accommodate technological change while avoiding the far-reaching policy consequences of recognising AI systems as authors or granting them legal personality through administrative interpretation.

As AI systems become more autonomous, questions of ownership, accountability and attribution will continue to evolve. For now, however, the position under Indian law appears clear, AI may be a powerful creative tool, but authorship remains anchored to a legally recognised human or juristic person.

Please follow and like us: