India’s Copyright Office Rules on AI Authorship: What the DABUS Order Actually Decided

On 31 August 2026, India's Registrar of Copyrights refused to register an artwork that named an AI system as its author. Most coverage summarised the outcome as "AI cannot be an author". The order says more than that. It names the human who counts, and it leaves a key question open.

The application

Dr Stephen Thaler applied in May 2022 to register “A Recent Entrance to Paradise”, an image generated by his AI system, DABUS.¹ The application named DABUS as author and Dr Thaler as owner.² Practitioners in the US and UK will know the name: the same work and the same claim to machine authorship have been tested in other jurisdictions.

According to Dr Thaler, he supplied photographs he had taken, curated thesaurus entries and sentences, and wrote short descriptions linking the two. The final image was then generated “without real-time human intervention” and “without any text prompt”.³

The Registrar appointed a Senior Advocate as amicus curiae⁴ and heard the matter three times. At the last hearing Dr Thaler was offered the chance to name himself as author instead. He declined.⁵

What the Registrar decided

India’s Copyright Act contains a provision most jurisdictions lack. Section 2(d)(vi) provides that the author of a computer-generated literary, dramatic, musical or artistic work is “the person who causes the work to be created”.⁶ The order turns on that phrase.

First, the machine is not the author. The Registrar held that DABUS “is not a natural or juristic person recognised in law and cannot be entered as author under Section 2(d)(vi)”.⁷ Strictly, this was not the decisive step. The order says that once DABUS is found not to have caused the work, whether it is a “person” “becomes academic and need not be determined”.⁸

Second, the test is causation, not generation. The Registrar held that the provision’s focus “is not upon the ‘entity that immediately generates the output’ but upon ‘the person who causes the work to be created’”.⁹ Section 2(d)(vi) “adopts a causation-based standard for the determination of authorship, and does not adopt a generation-based or an immediate-creation-based standard.”¹⁰ In the order’s words, “Autonomy in execution is not synonymous with conception of a work.”¹¹

Third, the human connection must be to the particular work. This is the point most summaries miss. “A person does not become the author of every output merely because that person owns, designed or developed the computer system.”¹² The order asks who exercised creative control and made “proximate and material arrangements” for the specific work.¹³ On Dr Thaler’s own account, it found that he “is the person who caused this particular work to be created within Section 2(d)(vi), although DABUS performed the immediate computational generation of its final visual form.”¹⁴

Fourth, owning the machine does not give you the copyright. Indian law makes the author the first owner. A claim to own a work whose named author cannot hold rights failed for want of any statutory route by which title could pass.¹⁵ “Ownership of a computer, software or other instrument does not, without more”, confer copyright in its outputs.¹⁶

The application was rejected because Dr Thaler maintained DABUS as author despite the opportunity to amend.¹⁷ Whether autonomous AI should ever be an author, the Registrar said, “remains a policy decision strictly reserved for Parliament”.¹⁸

What it did not settle

The Registrar also found that the image was original, so it met the Act’s originality requirement.¹⁹ “A work is not deprived of originality merely because technology played an indispensable role in its production.”²⁰ Read quickly, this suggests India protects wholly machine-generated output.

Three cautions apply:

• The order limits its own finding. The originality finding is made “for the limited purpose of the present registration proceedings”.²¹
• The result did not depend on it. The application failed on authorship and ownership, so the originality conclusion was not needed for the outcome. That is our reading, not the order’s words.
• The order keeps the question open. It does not attempt to rule on every form of AI-assisted creation, because “the degree and character of human involvement may vary materially”.²²

So the question that matters most to commercial users is still open in India: when someone generates content with a tool built by someone else, does that person have to show their own creative skill for copyright to exist?

How India now differs

In the United States, Dr Thaler’s argument that he was the author “by virtue of making and using” his machine was never decided. The D.C. Circuit held that the argument had been waived.²³ India’s Registrar reached that question and answered it in Dr Thaler’s favour on the facts, even while rejecting the application as filed.

The UK Government’s March 2026 report noted that India is one of a few countries with a computer-generated-works provision similar to the UK’s, and said that legislation in these countries “has been largely untested in the courts”.²⁴ The DABUS order is the first detailed Indian reading of that provision. It is still an administrative order of the Registrar, not a court judgment, and a court may take a different view.

Practical points for foreign counsel

• Name a legal person as author in Indian filings. An application naming an AI system as author will be refused. The Registrar will not rewrite the particulars for you.²⁵
• Keep records specific to each work. Under the order’s test, authorship follows the person’s acts for each work, not ownership of the tool. Record who supplied the inputs, configured the process and selected the output.
• Don’t assume protection for low-input outputs. The order’s originality finding is provisional, and the human-skill question is unresolved.

Footnotes

1. paras 1, 3 ↩

2. para 2(c); para 141 ↩

3. para 16 ↩

4. para 12 ↩

5. para 10 ↩

6. Copyright Act 1957, s.2(d)(vi), inserted by Copyright (Amendment) Act 1994 (38 of 1994), s.2, w.e.f. 10 May 1995 ↩

7. para 189(ii) ↩

8. para 128 ↩

9. para 93 ↩

10. para 114 ↩

11. para 111 ↩

12. para 98 ↩

13. para 127 ↩

14. para 107 ↩

15. Copyright Act s.17; paras 164, 170 ↩

16. para 165 ↩

17. paras 10, 190 ↩

18. para 189(vi) ↩

19. paras 48, 51 ↩

20. para 49 ↩

21. para 48; see also para 50 ↩

22. para 23 ↩

23. Thaler v Perlmutter, No. 23-5233 (D.C. Cir. 18 Mar 2025), slip op. 3 ↩

24. DSIT/DCMS, Report on Copyright and Artificial Intelligence (Mar 2026), Section I, para 16, p.106 ↩

25. paras 187, 190 ↩