1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

TEAM 46_R

1st CNLU-DPIIT-IPR NATIONAL MOOT COURT COMPETITION, 2026

BEFORE THE HON’BLE HIGH COURT OF INDRAPRASTHA

IP DIVISION

UNDER §20 OF THE CODE OF CIVIL PROCEDURE, 1908 AND §7 OF THE

COMMERCIAL COURTS ACT, 2015

IN THE MATTER OF

SHREYA SHEERAN......... PLAINTIFF

Versus

ED UDHAS AND OTHERS……… DEFENDANTS

MEMORIAL FOR THE DEFENDANTS

Page | I
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

TABLE OF CONTENTS

TABLE OF CONTENTS .............................................................................................. II

LIST OF ABBREVIATIONS ......................................................................................IV

LIST OF REFERENCES .............................................................................................. V

STATEMENT OF JURISDICTION .............................................................................. 1

STATEMENT OF FACTS ............................................................................................. 2

STATEMENT OF ISSUES ............................................................................................ 3

SUMMARY OF ARGUMENTS ................................................................................... 4

ARGUMENTS ADVANCED ........................................................................................ 6

1. The Plaintiff is not entitled to an interim injunction restraining the Defendants

from using any voice/video resembling her voice/performance pending disposal of the
suit 6

1.1 No Prima Facie case exists in the Plaintiff’s favour ...................................... 6

1.2 No irreparable injury has occurred to the Plaintiffs ....................................... 8

1.3. The Balance of Convenience exists in favour of the Defendants .................. 8

2. The Use of Shreya Sheeran’s Publicly Accessible Performances to Train the

Synthetic Voice Featured in “Tere Bin Main Kya” does not Amount to Infringement
under the Act .................................................................................................................. 9

2.1 The usage of performance recordings for training and output generation does
not amount to copyright infringement ..................................................................... 10

2.1.1 The use of Plaintiff’s performances for training SwaRaj does not amount
to copyright infringement .................................................................................... 11

2.2 The generation of Tere Bin Main Kya does not amount to copyright
infringement ......................................................................................................... 12

2.3 The usage is protected by the Fair Dealing exception ................................. 14

3. The Personality Rights of the Plaintiff have not been Violated ........................... 17

Page | II
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

4. The Defendants cannot be compelled to disclose their AI training datasets, and

model documentation ................................................................................................... 20

4.1 Compelled disclosure of trade secrets is disproportionate and causes

irreparable harm ....................................................................................................... 20

4.2 Requirements of relevance and necessity of disclosure are not met ............ 21

4.3 Availability of alternatives defeats the need for compelled disclosure ........ 22

5. The Defendants are not Liable to Pay Royalty to the Plaintiff ............................ 23

6. The Plaintiff is not entitled to permanent injunction restraining the Defendants

from using any voice or video resembling her voice or performances, or using any
datasets containing her performances .......................................................................... 25

6.1 Plaintiff’s rights are not undisputed to be entitled to a permanent injunction

26

6.2 Blanket restraint is overbroad, vague and legally impermissible ................ 26

6.3 No irreparable harm is established ............................................................... 28

6.4 Grant of permanent injunction would be detrimental to the economy ........ 28

PRAYER ...................................................................................................................... 30

Page | III
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

LIST OF ABBREVIATIONS

ABBREVIATION DEFINITION

& And

¶ Paragraph

etc. Etcetera

§ Section

Act The Copyright Act, 1957

AI Artificial Intelligence

Art. Article

DPIIT Department for Promotion of Industry and Internal Trade

of India

EU The European Union

HC High Court

SC Supreme Court

US The United States

s Section

TDM Text and Data Mining

WIPO World Intellectual Property Organization

v. Versus

Page | IV
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

LIST OF REFERENCES

S. NO. ARTICLES

1. Abraham Drassinower, ‘What’s Wrong with Copying’ (Harvard University

Press, 2015) 88-100.

2. Ann Cheryl Luk, ‘Analysing the Legal Meaning of ‘Copying’ through an

Empirical Study of UK Substantial Similarity Copyright Cases and
Psychological Explanations Of Litigant Trends’ (2025) 20(7) Journal of
Intellectual Property Law & Practice accessed 25 December 2025.

3. Benjamin L.W. Sobel, ‘Elements of Style: Copyright, Similarity, And

Generative AI’ (2024) 38(1) Harvard Journal of Law and Technology
accessed 24 December 2025.

4. Bertin Martens, ‘Economic Arguments in Favour of Reducing Copyright

Protection for Generative AI Inputs and Outputs’ (2024) Bruegel Working
Paper 09/2024 accessed 27 December 2025.

5. CISAC, ’ Global Creators and Performers Demand Creative Rights in AI

Proliferation’ (CISAC, 20 July 2023) <
[Link]
performers-demand-creative-rights-ai proliferation> accessed 22
December 2025.

6. Deepa Seetharaman, The Next Great Leap in AI Is Behind Schedule and

Crazy Expensive, Wall St. J. (Dec. 20, 2024), accessed 25 December 2025.

7. Directive (EU) 2019/790 of the European Parliament and of the Council of

17 April 2019 on copyright and related rights in the Digital Single Market
and amending Directives 96/9/EC and 2001/29/EC [2019] OJ L 130/92, art
3.

Page | V
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

8. Dorien Herremans, ‘Royalties in the Age of AI: Paying Artists for AI-
Generated Songs’ (WIPO Magazine, 6 May 2025) accessed 28 December
2025.

9. DPIIT, Balancing AI and Innovation (Working Paper on Generative AI and

Copyright Part1) accessed 21 December 2025.

10. Espie Angelica A. de Leon ‘Bollywood wants to be part of OpenAI India

copyright case’ (AsiaIP Law, 9 March 2025) accessed 22 December 2025.

11. Jenny Quang, ‘Does Training AI Violate Copyright Law’ (2021) 36 Berkey
Technology Law Journal accessed 24 December 2025.

12. Jessica L. Gillotte, ‘Copyright Infringement in AI-Generated Artworks’

(2020) 53 U.C. DAVIS L. REV. accessed 27 December 2025.

13. Mark A. Lemley and Bryan Casey, ‘Fair Learning’ (2021) 99(4) Texas Law
Review accessed 22 December 2025.

14. Mark A. Lemley, ‘How Generative AI Turns Copyright Upside Down’

(2024) 25 Science & Technology Law Review accessed 27 December 2025.

15. Matthe Sag, ‘Copyright Safety for Generative AI’ (2023) 61(2) Houston
Law Review accessed 23 December 2025.

16. McKinsey & Company, ‘The economic potential of generative AI: The next
productivity frontier’,(McKinsey & Company, 14 June 2023) accessed on
28 December 2025.

17. Ms. Astha Ojha, ‘AI & Copyright in India: Law, Policy, and the Future of
Creative Rights’ (NeGD, 8 October 2025) accessed on 29 December 2025.

18. Open Evidence v. Pathway: the Legal Battle Over AI Reverse Engineering
(Rutgers Law School¸4 March 2025) accessed 25 December 2025.

Page | VI
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

19. Pierre N. Leval, ‘Toward a Fair Use Standard’ (1990) 103(5) HARV. L.
REV. 1105, 1111.

20. Pragya Jha & Bernd Justin Jutte, ‘Does Human Learning equal Machine
Learning? High Court of Delhi to rule on lawfulness of TDM for Machine
Learning’ (Kluwer Copyright Blog, 19 May 2025) accessed 25 December
2025.

21. Shama Mahajan, ‘[Part I] ANI v. Open AI – A Lesson in Resisting the

Temptation to Borrow Excessively without Legislative Sanction’ (Spicy IP,
30 May 2025) accessed 24 December 2025. 18 R.G. Anand v. Deluxe Films
(1978) 4 SCC 118.

22. Sri. G. Shiva Prasad Yadav, ‘Declaration and Injunction Suits’ (Kadapa
District Court, 14 March 2024) accessed 20 December 2025.

23. Super Cassettes Industries Ltd v Hamar Television Network (P) Ltd 2010
SCC OnLine Del 2086.

24. Tusha Malhotra and Tanvi Bhatnagar, ‘Balancing individual liberties and
commercial exploitation: Indian personality rights jurisprudence’ (Anand
and Anand, 13 September 2024) accessed 24 December 2025.

25. UNCTAD, ‘Creative Economy Outlook 2024’ (UNCTAD, 11 July 2024)

accessed on 29 December 2025.

26. WIPO ‘Trade Secrets’ (WIPO) accessed 25 December 2025.

27. Yogesh Pai, ‘Patent Injunction Heuristics in India’ (2019) Kluwer Law
International BV accessed 25 December 2025.

BOOKS

28. H.K. Saharay (ed), Iyengar’s Commentary on the Copyright Act (9th edn,
Lexis Nexis 2016) 171.

Page | VII
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

INTERNATIONAL CASES

29. Authors Guild Inc. v. Google Inc., Case No. 13-4829-cv (2d Cir. 2015).

30. Baker v. Selden, 101 U.S. 99 (1879).

31. Bartz v. Anthropic PBC, Case No. 3:24-cv-05417, (N.D. Cal. 2024).

32. Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569 (1994).

33. Case C-5/08 Infopaq International A/S v. Danske Dagblades Forening

(2009), ECLI:EU:C:2009:465.

34. Centurion Indus. v. Warren Steurer and Assoc., 665 F.2d 323 (10th Cir.
1981).

35. Chancellor Masters and Scholars of the University of Oxford v. Narendra

Publishing House, (2008) 38 PTC 385.

36. Congoo, LLC v. Revcontent LLC, 2017 WL 3584205 (D.N.J. 2017).

37. GEMA v. Open AI, Higher Regional Court of Munich I 42 O 14139/24.

38. Generac Power Sys. Inc. v. Kohler Co., No. 11-1120 (E.D. Wisc. Jun. 6,
2012) (ECF No. 32).

39. Hubbard v. Vosper [1972] 2 Q.B. 84

40. MagicJack Vocaltec Ltd. v. [Link], Inc., No. 12-80360 (S.D. Fla. Oct.
18, 2012) (ECF No. 51).

41. Richard Kadrey, et al., v. Metal Platofrms, Inc. Case No. 23-cv-03417-VC
(N.D. Cal. 2025).

42. Sega Enterprises v. Accolade, Inc., Case No. 977 F.2d (9th Cir, 1992).

Page | VIII
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

43. Telspace, LLC v. Coast to Coast Cellular, Inc., 2014 WL 4364851, *4 (W.D.
Wash. Sept. 3, 2014).

INDIAN CASES

44. Amitabh Bachchan v Rajat Nagi, (2022) 6 HCC (Del) 641.

45. Anathula Sudhakar v P. Buchi Reddy (2008) 4 SCC 594.

46. Anil Kapoor v. Simply Life India Delhi High Court CS(COMM) 652/2023.

47. Arijit Singh v. Codible Ventures LLP 2024 SCC OnLine Bom 2445.

48. Asha Bhosle v. Mayk Inc 2025 SCC OnLine Bom 3485.

49. Associated Electronics & Electrical Industries v. Sharp Tools, AIR 1991
Kant 406.

50. Civic Chandran v. Amini Amma 1996 SCC OnLine Ker 63.

51. Dalpat Kumar v. Prahlad Singh, (1992) 1 SCC 719.

52. F-Hoffmann-La Roche AG & Anr. v Zydus Lifesciences Ltd 2025 SCC
OnLine Del 2347.

53. Gujarat Bottling Co. Ltd. v. Coca Cola Co, (1995) 5 SCC 545.

54. ICC Development (International) Ltd. v. Arvee Enterprises 2003 SCC

OnLine Del 2.

55. Indian Singers Rights Association v. Night Fever Club & Lounge 2016 SCC
OnLine Del 5418.

56. Myspace Inc. v. Super Cassettes Industries Ltd. 2016 SCC OnLine Del
6382.

57. R.G. Anand v. Deluxe Films, (1978) 4 SCC 118.

Page | IX
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

58. R. Rajagopal v. State of T.N. (1994) 6 SCC 632.

59. Shree Vardhaman Rice and General Mills v. Amar Singh Chawalwala
(2009) 10 SCC 257.

60. Sun Pharmaceutical Industries Ltd. v. Meghmani Lifesciences Ltd., Interim

Application (L) No. 9484 of 2025.

61. Syndicate of the Press of the University of Cambridge v. B.D. Bhandari &
Anr. 2011 SCC OnLine Del 3215.

62. Tips Industry Ltd. v. Wynk Ltd. & Anr. 2019 SCC OnLine Bom 13087.

63. Titan Industries Ltd. v. Ramkumar Jewellers 2012 SCC OnLine Del 282.

64. Transformative Learning Solutions (P) Ltd. v. Pawajot Kaur Baweja, 2019
SCC OnLine Del 9229.

STATUTES AND GUIDELINES

65. The Copyright Act 1957.

66. The Copyright Act, Japan.

67. 17 U.S. Code, s 107.

Page | X
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

STATEMENT OF JURISDICTION

It is humbly submitted that the Defendants have invoked the jurisdiction of

the Hon’ble High Court of Indraprastha, IP Division in pursuance of §20 of
the Code of Civil Procedure, 1908 and §7 of the Commercial Courts Act,
2015.

20. Other suits to be instituted where defendants reside or cause of action

arises—

Subject to the limitations aforesaid, every suit shall be instituted in Court

within the local limits of whose jurisdiction—

(a) the defendant, or each of the defendants where there are more than one,
at the time of the commencement of the suit, actually and voluntarily
resides, or carries on business, or personally works for gain; or

(c) the cause of action, wholly or in part, arises.

7. Jurisdiction of Commercial Divisions of High Courts.—

All suits and applications relating to Commercial disputes of a Specified

Value filed in a High Court having ordinary original civil jurisdiction shall
be heard and disposed of by the Commercial Division of that High Court:

Provided that all suits and applications relating to commercial disputes,

stipulated by an Act to lie in a court not inferior to a District Court, and
filed or pending on the original side of the High Court, shall be heard and
disposed of by the Commercial Division of the High Court:

The Defendants set forth the Facts, Issues and Arguments in the present
case in the jurisdiction of the High Court of Indraprastha, IP Division.

Page | 1
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

STATEMENT OF FACTS

Ed Udhas, an engineer who is trained in Machine Learning and Deep Neural Networks,
is the founder of Sarvagya Studios. In 2024, he developed a prototype generative AI-
model which is trained on the publicly available performances of renowned singers and
secured a substantial investment of Rs. 1.2 crore for its further development. He
launched his music generation application “SwaRaj” in late 2025 as an AI-based
customisable music creation platform. Ed Udhas, SwaRaj, Sarvagya Studios and the
investors are collectively referred to as the Respondents. The Plaintiff, Shreya Sheeran
is a highly acclaimed playback singers of Sindia, with training in diverse vocal genres.
She has won multiple prestigious awards and also received a Grammy Awards
nomination for the Best Global Music Performance category. Her musical performances
comprise works specifically for cinematograph films as well as independent singles,
which have consistently emerged as chart-leaders and essentials for celebrations,
broadcasting programmes, and personal streaming playlists.

The launch event featured the release of the SwaRaj generated song Tere Bin Main Kya.
Expert reports and public opinion regarding the similarity of its synthetic voice artist
V-23-F with the Plaintiff are cited by her to allege that the Respondent has infringed
her copyright and performer’s rights by using her performances for training SwaRaj.
However, SwaRaj has been trained ethically on an open-source music database
comprising the techniques (non-expressive elements) used by renowned domestic and
global singers. The training involves transformation of the collected data into numerical
representation for processing by the computer systems to learn the general musical
styles and structures. The outputs generated are determined by weighted models and
random choice by the AI based on its learning experiences in response to the user
prompts. The Defendants have clearly disclosed that the voice in Tere Bin Main Kya is
synthetic and not associated with or replica of a particular human voice. There has been
no reproduction of any literary or musical melody present in the Plaintiff’s
performances in Tere Bin Main Kya and even the reports do not conclusively prove any
copying. The losses claimed by the Plaintiff have not been established conclusively as
the halting negotiations is only temporary in nature and there is no actual proof of
substitution of her works by AI generated outputs in the industry.

Page | 2
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

STATEMENT OF ISSUES

ISSUE I

Whether, at the interlocutory stage, the Plaintiff is entitled to an interim injunction

restraining the Defendants from using any voice/video resembling her
voice/performance pending disposal of the suit?

ISSUE II

Whether the use of AI to train a voice model on publicly accessible performances of

Shreya Sheeran (if proven) and generating the song “Tere Bin Main Kya” using AI
amounts to an infringement under the Copyright Act, 1957?

ISSUE III

Whether the alleged replication of Shreya Sheeran’s vocal identity and performance
style violates her personality rights under Sindian law?

ISSUE IV

Whether the Defendants can be compelled to disclose their AI training datasets and
model documentation, balancing copyright enforcement with trade-secret protection?

ISSUE V

Whether Ed Udhas and Sarvagya Studios are liable to pay royalty or compensation for
the alleged use of Shreya Sheeran’s performances for AI training?

ISSUE VI

Whether the Plaintiff is entitled to permanent injunction restraining the Defendants

from using: (i) any voice resembling her voice; (ii) any video resembling her
performances; or (iii) datasets allegedly containing her performances?

Page | 3
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

SUMMARY OF ARGUMENTS

ISSUE I

The Plaintiff is not entitled to an interim injunction. The grant of an interim injunction
is an equitable remedy, discretionary in nature, and must be exercised with extreme
caution. As established in judicial precedents, the Plaintiff must satisfy the triple test of
(A) Prima Facie Case, (B) Irreparable Injury, and (C) Balance of Convenience. The
Plaintiff fails on all three counts.

ISSUE II

The use of the Plaintiffs works among other musical performances for training the
SwRaj application does not amount to copyright infringement. During training, the
collected information is tokenised in numeric representations, therefore the storage
occurs in a format that is different from the underlying works. Further, the expressive
elements of the copyrighted work are not published or communicated to the public in a
suitable format to access the protected expression. The generation of output is
dependent on the weights associated with different works present in the dataset and the
specific prompts given by the users. There is no infringement in the present case
because the alleged similarity exists only in vocal techniques used commonly by female
South Asian singers as they are an essential part of their musical training. There is no
allegation or evidence of similarity between the musical and literary elements of Tere
Bin Main Kya and songs performed by humans, particularly the Plaintiff. Such use is
further protected by the Fair Dealing defence under Section 52 of the Act because it has
a transformative purpose- training AI which generates social utility. Market harm has
not been proved conclusively and the instances presented by the Plaintiff are merely
speculative in nature.

ISSUE III

Even though voice is a protected attribute of a person’s identity under their personality
rights, the Plaintiff’s personality rights have not been violated in the present case. The
similarity exists only with regards to the common techniques used by female South

Page | 4
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

Asian singers. There is no harm to the exclusive performance rights or the moral rights
of the Plaintiff because none of her performances have been communicated to the public
or distorted by the generation of Tere Bin Main Kya. Further, it has been categorically
disclosed that the song features a synthetic voice named V-23-F, therefore, the
Defendants have not misappropriated the Plaintiff’s identity for their commercial
benefit.

ISSUE IV

The Defendants cannot be compelled to disclose their AI training datasets and model
documentation because, firstly, compelled disclosure of a trade secret is a
disproportionate step and would cause irreparable harm, secondly, Copyright Law does
not necessitate disclosure of datasets to establish infringement, and thirdly, there are
alternatives available, which defeats the need for compelled disclosure.

ISSUE V

The Plaintiff is not entitled to receive reasonable royalties. Royalties refer to the
payment of the pre-agreed percentage of sales or profits achieved by the licensee to the
licensor for the use of such work. The right to royalty is a statutory creation, not a
common law right. Under the Copyright Act, royalties are payable only when a work is
utilized in a manner specifically prescribed under Section 38A (e.g., reproduction,
broadcast, or communication to the public

ISSUE VI

The Plaintiff is not entitled to any permanent injunction restraining the Defendants.
Firstly, no prima facie case is made out in favour of the Plaintiff. Secondly, a blanket
restraint is overbroad, vague and legally impermissible. Thirdly, no irreparable harm is
established and a permanent injunction would stifle innovation.

Page | 5
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

ARGUMENTS ADVANCED

1. THE PLAINTIFF IS NOT ENTITLED TO AN INTERIM INJUNCTION RESTRAINING THE

DEFENDANTS FROM USING ANY VOICE/VIDEO RESEMBLING HER

VOICE/PERFORMANCE PENDING DISPOSAL OF THE SUIT

1. It is submitted that the Plaintiff is not entitled to the grant of an interim injunction.
An injunction is an equitable and discretionary relief, which must be exercised
sparingly and only where the Plaintiff strictly satisfies the settled three-fold test of
existence of a prima facie case (1.1), likelihood of irreparable injury (1.2), and
balance of convenience (1.3). Gujarat Bottling Co. Ltd. v. Coca Cola Co., held that
interim injunctions are not to be granted merely because a triable issue exists; the
Court must be satisfied that equity, justice, and convenience demand restraint.1

2. It is further submitted that in cases of copyright or performer’s rights infringement,

courts do not proceed on either isolated similarities or subjective impressions. The
settled position of law is that the impugned work must be examined as a whole, and
infringement can be found only where the allegedly copied material constitutes a
protectable part of the original work, resulting in prejudicial appropriation.2 Mere
stylistic resemblance, similarity in genre, or a general artistic impression is not
sufficient to establish a prima facie case.

3. In the present case, the Plaintiff seeks an extraordinarily broad restraint against the
use of any voice or video resembling her voice or performance, without
demonstrating the direct copying of identifiable performance, fixation, or recording
that is protected under the Copyright Act. If such an injunction is granted, it would
amount to conferring a monopoly over vocal style and performance aesthetics. This
protection is not a remedy at the interlocutory stage.

1.1 No Prima Facie case exists in the Plaintiff’s favour

1
Gujarat Bottling Co. Ltd. v. Coca Cola Co. (1995) 5 SCC 545.
2
Bikramjeet Singh Bhullar Vs Yash Raj Films Private Limited & Ors (2023) SCC OnLine Del 8212.

Page | 6
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

4. The Plaintiff has failed to establish a prima facie case of infringement or the
common law tort of passing off. Firstly, there is no copyright in vocal style or
method of singing. Indian Copyright law protects specific works (literary, musical,
sound recordings), not a style or method. RG Anand v. Delux Films,3 held there is
no copyright in an idea, theme, or style but only in the specific expression.

5. The Plaintiff’s claim over ‘breath-silk vibrato’ or ‘breathing intervals’ is an attempt

to monopolize a technique of singing. Such techniques are part of the common stock
of artistic skill and are incapable of exclusive ownership. The Supreme Court has
clarified that copyright does not protect mere skill, labour, or industry unless it
culminates in a distinct original expression fixed in a protectable work.4

6. The AI model V-23-F was trained on a dataset of female South Asian vocal styles5,
not exclusively on the Plaintiff. The resulting voice is a generic blend. In Raciraft
v. EsiRaft, the court vacated an injunction because visual or phonetic similarity
alone does not amount to passing off if the underlying elements (like a generic color
scheme or, in this case, a generic vocal style) are common to the trade.6

7. Secondly, this case must be seen as distinct from personality rights jurisprudence.
The plaintiffs may place reliance on precedents such as Amitabh Bachchan v. Rajat
Nagi & Ors.7 However, in those cases, the defendants explicitly used the names,
photos, and specific characters of the celebrities to sell products. In this case, there
is no use of the name of the artist, no photo of her, and no claim that she endorsed
the app. The confusion is subjective to the audience, and not induced by the
Defendants' misrepresentation.

8. Thirdly, Section 52(1) of the Copyright Act allows for fair dealing.8 The use of data
for machine learning is distinct from copying a song for sale and commercial
purposes. Defendants are democratizing the creation of music, as stated by Ed

3
R.G. Anand v. Deluxe Films (1978) 4 SCC 118.
4
Eastern Book Company v. D.B. Modak, 2008 1 SCC 1.
5
Moot Proposition, Annexure F.
6
Sun Pharmaceutical Industries Ltd. v. Meghmani Lifesciences Ltd., Interim Application (L) No. 9484
of 2025.
7
Amitabh Bachchan v Rajat Nagi (2022) 6 HCC (Del) 641.
8
The Copyright Act 1957, s 52.

Page | 7
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

Udhas.9 Restraining this innovation based on a subjective claim would stifle

technological progress in the name of IP protection, which is contrary to the intent
of the Copyright Act. Thus, there is no prima facie case in favour of the Plaintiff.

1.2 No irreparable injury has occurred to the Plaintiffs

9. The Defendants submit that the settled position of law is that where the injury
complained of is economic or commercial in nature, and can be quantified in
monetary terms, the grant of an interim injunction is unwarranted. The Hon’ble
Supreme Court has held that injunctions ought not to be granted where damages
constitute an adequate remedy.10

10. The Plaintiff cannot demonstrate that money is an inadequate remedy. If the
Plaintiff eventually succeeds in its claims, any loss of royalties or endorsement fees
can be calculated and compensated monetarily.

11. Moreover, the market dilution argument is merely speculative. The App targets
amateur creators, whereas the Plaintiff targets high-end film production and luxury
endorsements.11 These are different markets. A ‘generated’ song does not compete
with a live performance or a film playback by a well-known artist. The Plaintiff’s
career is built on her persona, which remains untouched. Courts have recognised
that where parties cater to different consumer bases and price points, claims of
direct competition and irreparable market harm are unsustainable.12

12. Accordingly, the Plaintiff has failed to demonstrate that the alleged injury is
irreparable in nature or that damages would be an inadequate remedy.

1.3. The Balance of Convenience exists in favour of the Defendants

13. Firstly, an injunction would require shutting down the core feature of the SwaRaj
app which is the V-23-F model. This would effectively halt the Defendants'

9
Moot Proposition, Annexure F.
10
Best Sellers Retail (India) Pvt. Ltd. v. Aditya Birla Nuvo Ltd (2012) 6 SCC 792; Wander Ltd. v. Antox
India Pvt. Ltd. (1990) Supp SCC 727.
11
Moot Proposition, ¶ 7.
12
Khoday Distilleries Ltd. v. Scotch Whisky Association (2008) 10 SCC 723.

Page | 8
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

business operations before a trial has even begun. Where an injunction would result
in irreversible harm to the defendant's business, the balance of convenience must
tilt against such restraint.13
14. Secondly, as noted in Dalpat Kumar v. Prahlad Singh14, the court must weigh the
mischief. Granting an injunction harms the public interest by denying access to
affordable music creation tools. It prioritizes the monopoly of one artist over the
creative expression of thousands of potential users.

15. Thirdly, the balance of convenience must also account for the reversibility of harm.
If the Defendants are permitted to continue operations pending trial, they can be
directed to maintain detailed accounts of revenues and profits, ensuring that any
eventual liability can be adequately compensated through damages or profit-
sharing. Conversely, if the application is injuncted at the early stage of their
business and the Defendants ultimately succeed, the loss of market momentum,
investor confidence, and first mover advantage would be permanent and irreparable.
Courts have recognised that loss of commercial opportunity and business viability
constitutes a grave prejudice.

16. Additionally, the Plaintiff’s requested restraint of prohibiting the use of any voice
resembling her voice or performance is vague and incapable of precise enforcement.
Granting such an overbroad injunction would expose the Defendants to uncertainty
in compliance and potential contempt. This would exacerbate hardship beyond what
equity permits at the interlocutory stage. Courts have consistently disfavoured
injunctions that are uncertain in scope or oppressive in effect. 15Therefore, the
balance of convenience is skewed in favour of the Defendants.

2. THE USE OF SHREYA SHEERAN’S PUBLICLY ACCESSIBLE PERFORMANCES TO

TRAIN THE SYNTHETIC VOICE FEATURED IN “TERE BIN MAIN KYA” DOES NOT
AMOUNT TO INFRINGEMENT UNDER THE ACT

13
Wander Ltd. v. Antox India Pvt. Ltd. (1990) Supp SCC 727.
14
Dalpat Kumar v. Prahlad Singh, (1992) 1 SCC 719.

15
Kewal Singh v. Lajwanti (1980) 1 SCC 290.

Page | 9
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

17. SwaRaj application has been developed by Ed Udhas’s AI-based musical startup
Sarvagya Studios. The Plaintiff has alleged that the AI generated song Tere Bin
Main Kya is similar to her vocal style and that the accompanying visual elements
are identical to her live convert performance. The Defendants submit that the usage
of publicly available musical performances and other sound samples for training AI
systems based on which new outputs are generated autonomously does not
constitute copyright infringement; (2.1) and such usage is protected by the fair
dealing defence under Section 52 of the Act. (2.2)

2.1 The usage of performance recordings for training and output generation
does not amount to copyright infringement

18. Copyright is the exclusive right of the owner to reproduce, store, translate, adapt,
communicate their work to the public, etc. according to Section 14.16 Original
literary, dramatic, musical, and artistic works along with sound recordings are
protected by copyright under Section 13(1).17 Musical works combine the lyrics and
composition of the song and are protectable by copyright as a separate category
from sound recordings.18 Further, a performer is granted special rights including
exclusive rights for making, reproducing, and communicating sound or visual
recording of their performance and protecting their moral rights under Section 38.19

19. It is submitted that it is not the case of the Defendants that performers and their
works should not be protected by the copyright regime. It is the case of the
Defendants that performance techniques constitute ideas, which are not protected
by copyright under the classic idea-expression dichotomy under the copyright law.20

20. It is submitted that generative AI systems consist neural networks to identify

patterns in the input dataset, understand and synthesise the data to generate output
tailored to the prompts submitted by users. The operation of AI-music creation

16
The Copyright Act 1957, s 14.
17
The Copyright Act 1957, s 13(1).
18
Espie Angelica A. de Leon ‘Bollywood wants to be part of OpenAI India copyright case’ (AsiaIP Law,
9 March 2025) <[Link]
copyright-case> accessed 22 December 2025.
19
The Copyright Act 1957, s 38.
20
Baker v. Selden, 101 U.S. 99 (1879).

Page | 10
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

platforms can be divided into two stages: (a) training it using datasets comprising
publicly available musical performances, sound recordings, conversations, etc. and
(b) generation of new outputs based on their ‘learning’ from the datasets.21

21. It is humbly submitted that the usage of publicly available performances, including
those of Shreya Sheeran for training SwaRaj (2.1.1) and the generation of Tere Bin
Main Kya by the synthetic voice V-23-F (2.1.2) does not constitute infringement.

2.1.1 The use of Plaintiff’s performances for training SwaRaj does not amount to
copyright infringement

22. The idea-expression dichotomy grants copyright protection only to the expressive
elements of a work.22 The functional and technical ideas and processes contained in
the are not protected.23 Mere technical or non-communicative uses of a copyrighted
work for non-expressive purposes such as training an AI model do not infringe the
copyright protection.

23. The AI training phase constitutes three stages, (a) collection of raw data; (b)
tokenisation of the collected data; (c) training of the model using the tokenised
datasets.24 During tokenisation, the copyrighted data scraped in the collection phase
is transformed into mathematical (numeric) tokens because computer systems can
process and synthesise information only when it is represented in the binary
format.25 Therefore, the expressive form of the copyrighted works stored in the
datasets is different from that of the original work.

21
Shama Mahajan, ‘[Part I] ANI v. Open AI – A Lesson in Resisting the Temptation to Borrow
Excessively without Legislative Sanction’ (Spicy IP, 30 May 2025) <[Link]
ani-v-open-ai-a-lesson-in-resisting-the-temptation-to-borrow-excessively-without-legislative-
[Link]> accessed 24 December 2025.
22
R.G. Anand v. Deluxe Films (1978) 4 SCC 118.
23
Abraham Drassinower, ‘What’s Wrong with Copying’ (Harvard University Press, 2015) 88-100.
24
Pragya Jha & Bernd Justin Jutte, ‘Does Human Learning equal Machine Learning? High Court of
Delhi to rule on lawfulness of TDM for Machine Learning’ (Kluwer Copyright Blog, 19 May 2025)
<[Link]
high-court-of-delhi-to-rule-on-lawfulness-of-tdm-for-machine-learning/> accessed 25 December 2025.
25
Shama Mahajan, ‘[Part I] ANI v. Open AI – A Lesson in Resisting the Temptation to Borrow
Excessively without Legislative Sanction’ (Spicy IP, 30 May 2025) <[Link]
ani-v-open-ai-a-lesson-in-resisting-the-temptation-to-borrow-excessively-without-legislative-
[Link]> accessed 24 December 2025.

Page | 11
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

24. Tips Industry v. Wynk Ltd., held that copyright infringement occurs upon
reproduction of the expressive elements of an original work combined with its
publication in a form accessible to the public for the usage of the expression.26
Reproduction of an original work, including its electronic storage, is an exclusive
right of the copyright holder under Section 14(a)(i).27 However, the numerical
representation does not replicate the original expression of the work and therefore
does not constitute reproduction of the copyright protected expression.

25. The copying of works is undertaken only to access the underlying ideas, facts,
linguistic and artistic structures of the works which are uncopyrightable. Such
copying is fundamental to the learning phase of generative AI models.28
Furthermore, the tokenised format of the data is only used for training AI models
and the expressive elements of the original works are not redistributed or
communicated to the public.29 Therefore, such training does not constitute
infringement of the copyright protection.

26. It is submitted that in the present case, SwaRaj application is trained ethically on
open-source machine learning models.30 The training involves copying of data in
non-expressive format and the dataset is not communicated to the public except in
machine readable codes which are different from the original expression of musical
works included in the dataset. Therefore, the present storage in numerical
representation after tokenisation and which does not allow public access to the
expressive elements of the copyrighted works does not amount to reproduction.

2.2 The generation of Tere Bin Main Kya does not amount to copyright infringement

27. It is submitted that the generation of outputs in response to a request involves

random choice among the possible permutations and combinations of the elements

26
Tips Industry Ltd. v. Wynk Ltd. & Anr. 2019 SCC OnLine Bom 13087.
27
The Copyright Act 1957, s 14(a)(i).
28
Mark A. Lemley and Bryan Casey, ‘Fair Learning’ (2021) 99(4) Texas Law Review
<[Link] accessed 22 December 2025.
29
Jenny Quang, ‘Does Training AI Violate Copyright Law’ (2021) 36 Berkey Technology Law Journal
<[Link] accessed 24 December 2025.
30
Moot Proposition, ¶ 7.

Page | 12
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

included in dataset.31 In order to constitute infringement, the output generated must

bear substantial similarity with the copyrighted works comprising the datasets due
to memorisation and regurgitation of their expressive elements.32

28. In GEMA v. OpenAI, the Regional Court of Munich decided that recognisable
reproduction of lyrics in AI generated outputs violated the copyright and was not
protectable under the EU text ad data mining exceptions.33 However, it is submitted
that in the present case, the song Tere Bin Main Kya is a completely new song
generated by SwaRaj34 and featuring the synthetic voice V-23-F.35 There is no
copying of any copyright-protected lyrics or composition (rhythmic arrangements)
from any pre-existing musical work within and beyond the dataset.

29. It is submitted that the artificial voice has been trained on an expansive dataset
comprising different female South Asian vocal styles and is not a replication of the
voice of any individual singer.36 Furthermore, vibrato, breathing intervals,
microtonal embellishments such as murki, etc. are not protectable by copyright
because they are universal musical techniques, i.e. ideas.37

30. It is submitted that there is no reproduction or redistribution of any musical

performance or song featuring Shreya Sheeran by the generation of Tere Bin Main
Kya. The waveform analysis does not prove the replication of lyrics (literary work)
and composition (musical work), and similarity in techniques do not prove
infringement.38

31
Mark A. Lemley, ‘How Generative AI Turns Copyright Upside Down’ (2024) 25 Science &
Technology Law Review <[Link]
[Link]> accessed 27 December 2025.
32
Matthe Sag, ‘Copyright Safety for Generative AI’ (2023) 61(2) Houston Law Review
<[Link] accessed 23 December 2025.
33
GEMA v. Open AI, Higher Regional Court of Munich I 42 O 14139/24.
34
Moot Proposition, ¶ 9.
35
Moot Proposition, Annexure F.
36
Ibid.
37
Mark A. Lemley, ‘How Generative AI Turns Copyright Upside Down’ (2024) 25 Science &
Technology Law Review <[Link]
[Link]> accessed 27 December 2025.
38
Moot Proposition, ¶ 19.

Page | 13
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

31. Further, the video generated by SwaRaj does not reproduce Shreya’s physical
likeness,39 and adheres to the standard visual style and aesthetical patterns adopted
by the music industry. Thus, the generation of Tere Bin Main Kya does not constitute
infringement under the Act because it is a new musical work with similarity only in
universal techniques and no copying of protected literary and musical works.

2.3 The usage is protected by the Fair Dealing exception

32. In light of the increasing importance of ethically trained AI in economic

development, governance and decision-making, various international jurisdictions
have enacted Text and Data Mining (TDM) exceptions for the lawful scrapping of
information to create improved datasets. The EU grants mandatory TDM exception
for scientific purposes40 and general TDM exception for commercial purposes while
providing machine readable ‘opt-out’ option to rights holders.41

33. Similar exceptions are available in Japan, subject to competition among the
copyrighted work and the AI generated output and impediments created by the latter
for the future sales of the former.42 However, such exceptions are not yet been
included in the Indian Copyright law, which only provides a fair dealing protection
for private use, research, criticism, review under Section 52(1)(a) of the Act.43

34. The US does not provide a specific exemption for TDM activities and relies on the
four-factor test to determine fair use,44 as applied in Campbell v. Acuff-Rose
Music.45 The test was adopted by the Indian legal system in Chancellor Masters of
the University of Oxford v. Narendra Publishing House,46 to determine the
constituents of fair dealing. The test considers the purpose and character of the use,

39
Moot Proposition, ¶ 21.
40
Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright
and related rights in the Digital Single Market and amending Directives 96/9/EC and 2001/29/EC [2019]
OJ L 130/92, art 3.
41
Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright
and related rights in the Digital Single Market and amending Directives 96/9/EC and 2001/29/EC [2019]
OJ L 130/92, art 4.
42
The Copyright Act, Japan, s 3(3).
43
The Copyright Act 1957, s 52(1)(a).
44
17 U.S. Code, s 107.
45
Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569 (1994).
46
Chancellor Masters and Scholars of the University of Oxford v. Narendra Publishing House, (2008)
38 PTC 385.

Page | 14
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

the nature of the copyrighted work, the amount and substantiality of the impugned
portion in relation to the copyrighted work as a whole, and the effect of the use upon
the potential market or the values of the copyrighted work.

35. It is submitted that transformative use of the original data for the creation of new
information, insights and aesthetics is a central enquiry while assessing the first
factor of the test.47 In BD Bhandari, it was held that the purpose and character of
the impugned work and the copyrighted work must be different; the former should
be transformative and not a mere substitute for the latter to avail protection under
fair dealing exception.48

36. In Authors Guild v. Google, the copying of textbooks for making them searchable
on the google database was held to constitute transformative use for its socially
beneficial purposes. The usage of snippets for providing facts and enhancing public
accessibility was protected because it served a functional purpose and provided
social utility.49 The purpose of using publicly available data is to develop AI systems
capable of creating new and contextually appropriate content across settings by
learning the underlying features and patterns present in the dataset and not to
memorise the elements of original expression of the individual works,50 and is
highly transformative.51

37. It is submitted that the use of copyrighted musical performances contained in open-
source machine learning datasets to train SwaRaj is transformative. It is
substantially different from the original use of musical performances, such as
enjoyment. SwaRaj is a customisable music creation platform which allows its users
to generate original composition, create unique musical styles, choose their

47
Pierre N. Leval, ‘Toward a Fair Use Standard’ (1990) 103(5) HARV. L. REV. 1105, 1111.
48
Syndicate of the Press of the University of Cambridge v. B.D. Bhandari & Anr. 2011 SCC OnLine Del
3215.
49
Authors Guild Inc. v. Google Inc., Case No. 13-4829-cv (2d Cir. 2015).
50
DPIIT, Balancing AI and Innovation (Working Paper on Generative AI and Copyright Part1)
<[Link] accessed
21 December 2025.
51
Jessica L. Gillotte, ‘Copyright Infringement in AI-Generated Artworks’ (2020) 53 U.C. DAVIS L.
REV. <[Link]
5_Gillotte.pdf> accessed 27 December 2025.

Page | 15
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

preferred aesthetic setup for their compositions and create personalised music.52
The outputs generated are unique to the choice of prompts and the selection of
available options. These democratise the capability to generate music to individuals
who were unable to harness their artistic creativity owing to lack of economic
resources and other opportunities,53 and serve a social purpose.

38. It is submitted that the nature of the copied work is incidental and necessary to the
training process.54 It is an intermediate step to access the unprotectable ideas and
functional elements55 such as the musical techniques (vibrato, breathing styles and
microtonal embellishments). The non-expressive nature of copying combined with
the non-publication of the expressive elements of the original dataset weigh in the
Defendants’ favour, regardless of the amount of copied material.56

39. In Andrea Bartz v. Anthropic PBC, the contention regarding market dilution owing
to competition between AI generated output and the works used for training was
rejected because the copyright law is not concerned with such creative displacement
and protection of authors from competition.57 In Kadrey v. Meta, it was held that it
is necessary for the plaintiffs to provide sufficient empirical evidence to establish
harm caused by the generated output to the core market of the copyrighted works.58

40. It is submitted that in the present case, the Plaintiff has failed to sufficiently
establish the economic or reputational harm arising from the generation of the song
Tere Bin Main Kya. The negotiations for an endorsement contract with a luxury
brand have only been paused temporarily, and neither the producers nor the
audience have actually substituted SwaRaj generated musical creations for the
Plaintiff’s musical performances.59 Furthermore, the decision to withhold the

52
Moot Proposition, ¶ 8.
53
Moot Proposition, Annexure F.
54
Bertin Martens, ‘Economic Arguments in Favour of Reducing Copyright Protection for Generative AI
Inputs and Outputs’ (2024) Bruegel Working Paper 09/2024
<[Link]
accessed 27 December 2025.
55
Sega Enterprises v. Accolade, Inc., Case No. 977 F.2d (9th Cir, 1992).
56
Mark A. Lemley and Bryan Casey, ‘Fair Learning’ (2021) 99(4) Texas Law Review
<[Link] accessed 22 December 2025.
57
Bartz v. Anthropic PBC, Case No. 3:24-cv-05417, (N.D. Cal. 2024).
58
Richard Kadrey, et al., v. Metal Platofrms, Inc. Case No. 23-cv-03417-VC (N.D. Cal. 2025).
59
Moot Proposition, ¶ 26.

Page | 16
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

release of a debut singer’s music track60 cannot be attributed to the launch of

SwaRaj AI due to its indirect and remote connection with the present dispute.

41. In Civic Chandran, the court relied on the decision in Hubbard v. Vosper,61 that fair
dealing is ultimately determined by the impression created on the viewers. 62 In the
present case, the impugned song does not reproduce any lyrics, melody or
composition arrangement that would create the impression of similarity with a
particular pre-existing musical performance of the Plaintiff. The similarity is merely
incidental and exists only for the common techniques63 adopted by female South
Asian singers and is not sufficient to prove infringement.

42. The transformative and non-expressive usage of the training dataset in the creation
of SwaRaj serves a social utility and the instances of economic and reputational loss
as contended by the Plaintiff are speculative in nature. Therefore, the usage of
publicly available performances to train SwaRaj and the generation of Tere Bin
Main Kya does not constitute infringement under the Act.

3. THE PERSONALITY RIGHTS OF THE PLAINTIFF HAVE NOT BEEN VIOLATED

43. It is submitted that the personality rights regime in Sindia is governed by the
fundamental right to privacy, the tort of passing off and performers’ exclusive and
moral rights under Section 38.64 In R. Rajagopal v. State of T.N., the right to control
the commercial use of identity was recognised as an element of the right to privacy
under Article 21.65 In Arvee Enterprises, attributes of an individual’s personality

60
Moot Proposition, ¶ 21.
61
Hubbard v. Vosper [1972] 2 Q.B. 84.
62
Civic Chandran v. Amini Amma 1996 SCC OnLine Ker 63.
63
Ann Cheryl Luk, ‘Analysing the Legal Meaning of ‘Copying’ through an Empirical Study of UK
Substantial Similarity Copyright Cases and Psychological Explanations Of Litigant Trends’ (2025) 20(7)
Journal of Intellectual Property Law & Practice <[Link] accessed 25
December 2025.
64
Tusha Malhotra and Tanvi Bhatnagar, ‘Balancing individual liberties and commercial exploitation:
Indian personality rights jurisprudence’ (Anand and Anand, 13 September 2024)
<[Link]
exploitation-indian-personality-rights-jurisprudence/> accessed 24 December 2025.
65
R. Rajagopal v. State of T.N. (1994) 6 SCC 632.

Page | 17
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

such as name, signature, etc. were held protectable under publicity rights,66 which
is the right to control the control the use of human identity as commercial assets.67

44. In Anil Kapoor v. Simply Life India, distinguishing elements of a celebrity’s persona
such as voice were held to be protected under personality rights.68 Indian courts in
Arijit Singh v Codilble Ventures LLP,69 and Asha Bhosle v. Mayk Inc.70 have held
that the utilisation of AI tools to create audio-visual content which reproduce
attributes identifiable with the performer, such as their voice, performance style,
aesthetic arrangements of their performance etc. amounts to unauthorised
commercial usage of their work and reputation.

45. However, it is submitted that in these cases, the AI tools were trained and designed
to mimic the voice of only a particular singer based on their performances and other
sound samples.71 Further, these cases involved the unauthorised use of their
personal attributes such as name, physical personality such as morphed pictures and
catchphrase dialogues to create the false impression of endorsement for the products
sold by the defendant AI system developers and website operators.72

46. It is further submitted that the present case is distinguished from the legal precedents
decided by the Indian courts because there is no misappropriation of Shreya’s
identity by using her name, physical likeness or vocal style to create a false
impression among audience for any endorsement. The Defendants have repeatedly
clarified that Tere Bin Main Kya is generated by AI and features a synthetic voice.73

66
ICC Development (International) Ltd. v. Arvee Enterprises 2003 SCC OnLine Del 2.
67
Titan Industries Ltd. v. Ramkumar Jewellers 2012 SCC OnLine Del 282.
68
Anil Kapoor v. Simply Life India Delhi High Court CS(COMM) 652/2023.
69
Arijit Singh v. Codible Ventures LLP 2024 SCC OnLine Bom 2445.
70
Asha Bhosle v. Mayk Inc 2025 SCC OnLine Bom 3485.
71
Tusha Malhotra and Tanvi Bhatnagar, ‘Balancing individual liberties and commercial exploitation:
Indian personality rights jurisprudence’ (Anand and Anand, 13 September 2024)
<[Link]
exploitation-indian-personality-rights-jurisprudence/> accessed 24 December 2025.
72
Supra note 57 and 58.
73
Moot Proposition, Annexure F.

Page | 18
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

47. Vocal style comprises techniques such as rhythm, pace, tonal modulation, etc. which
are capable of being learnt without replicating a particular singer.74 In the present
case, Shreya’s performances are part of a broader open-source dataset of musical
works. The synthetic voice V-23-F is trained using a blend of non-expressive
techniques and vocal methodologies used by numerous South Asian female singers,
and does not replicate one specific individual.75

48. It is submitted that Shreya’s exclusive performer rights under Section 38A of the
Act have not been violated by the video performance of Tere Bin Main Kya as it is
an entirely new creation of SwaRaj application and does not reproduce, copy,
communicate or monetarily profit from any of her pre-existing protected
performances.76 Further, her moral rights under Section 38B of the Act have not
been violated because there is no distortion, mutilation or modification of any of
her performances which is harmful to her reputation. 77

49. Furthermore, the aesthetic similarity is attributable to the standard lightning patterns
and visual styles used across the industry for live performances and do not harm
Shreya’s personality rights. The training of SwaRaj based on South Asian musical
techniques which are an essential part of the training of any South Asian singer
along with no similarity in lyrics or musical works to any of Shreya’s performances
proves that SwaRaj does not replicate her vocal identity.

50. It is further submitted that the constituents for passing off are not fulfilled in the
present case. The Defendants have unequivocally disclosed Tere Bin Mian Kya as
an AI generated song performed by a synthetic voice. Hence, the possibility of
misrepresentation and consequent damage due to confusion among audience does
not exist.78 Therefore, the personality rights of Shreya Sheeran have not been
violated under the Sindian legal regime.

74
Benjamin L.W. Sobel, ‘Elements of Style: Copyright, Similarity, And Generative AI’ (2024) 38(1)
Harvard Journal of Law and Technology <[Link] accessed 24 December
2025.
75
Moot Proposition, Annexure F.
76
The Copyright Act 1957, s 38A.
77
The Copyright Act 1957, s 38B.
78
Titan Industries Ltd. v. Ramkumar Jewellers 2012 SCC OnLine Del 282.

Page | 19
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

4. THE DEFENDANTS CANNOT BE COMPELLED TO DISCLOSE THEIR AI TRAINING

DATASETS, AND MODEL DOCUMENTATION

51. It is humbly submitted that the Defendants cannot be compelled to disclose their AI
training datasets and model documentation because, firstly, compelled disclosure of
a trade secret is a disproportionate step and would cause irreparable harm, (4.1)
secondly, Copyright Law does not necessitate disclosure of datasets to establish
infringement, (4.2) and thirdly, there are alternatives available, which defeats the
need for compelled disclosure.(4.3)

4.1 Compelled disclosure of trade secrets is disproportionate and causes

irreparable harm

52. Trade secrets encompass both technical and commercial information, which, is
confidential business information that provides an enterprise a competitive edge and
is unknown to others.79 The training dataset and documentation on which SwaRaj
is modelled, constitutes a trade secret, for same reasons, that it constitutes technical
information, giving the Defendants a competitive edge and is unknown to others.

53. Further, it is humbly submitted, that even though AI training datasets are usually
developed using publicly accessible audio recordings, concerts, interviews and
studio sessions, the same does not lead to its disqualification as a trade secret. This
is because, WIPO explicitly clarifies that a trade secret may be made up of a
combination of elements, each of which by itself is in the public domain, but where
the combination, which is kept secret, provides a competitive advantage.80

54. In Open Evidence Inc. v. Pathway Medical, Inc., Pathway allegedly tricked Open
Evidence’s platform into divulging portions of its system prompt, which is the
instruction set that governs how the AI model operates.81 This was alleged to

79
WIPO ‘Trade Secrets’ (WIPO) <[Link] accessed 25 December
2025.
80
Ibid.
81
Open Evidence v. Pathway: the Legal Battle Over AI Reverse Engineering (Rutgers Law School¸4
March 2025) <[Link]
reverse-engineering/> accessed 25 December 2025.

Page | 20
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

constitute a core trade secret akin to proprietary source code and, therefore, to trade
secret misappropriation, unfair competition, and breach of contract.82

55. That trade secret discovery causes irreparable harm is further strengthened by the
decision of Covey Oil Co. v. Continental Oil Co., wherein the Court held that in
trade secret discovery, the need for information in a case must be balanced against
the potential for irreparable competitive harm.

56. It is submitted, that companies have become increasingly protective of their systems
and breakthroughs. Competition amongst AI labs have grown so fiercely that major
tech companies publish fewer papers about their recent breakthroughs, because the
results of their research are now considered as trade secrets that need guarding.83

4.2 Requirements of relevance and necessity of disclosure are not met

57. To compel the production of source code, the plaintiff must prove that it is relevant
and necessary to the action. Reliance is placed on Centurion Indus v. Warren Steurer
and Assoc., which held that it is the discretion of the trial court to decide whether
trade secrets are relevant and whether the need outweighs the harm of disclosure.84

58. In MagicJack Vocaltec Ltd. v. [Link], Inc., the Court denied the motion to
compel source code because the plaintiff failed to persuade that the source code is
necessary to fully analyze the Defendants’ accused devices.85 Further, even though
there was a protective order, the court held that it did not provide enough protection,
and thus believed that disclosure could be fatal to the business of the defendant.

59. In Congoo, LLC v. Revcontent LLC, a US Court held that the plaintiff did not meet
the burden of demonstrating that production of source code is relevant and
necessary, and hence, denied the motion to compel the inspection and production of
the defendant’s source code.86 In that case, the defendants had claimed that

82
Ibid.
83
Deepa Seetharaman, The Next Great Leap in AI Is Behind Schedule and Crazy Expensive, Wall St. J.
(Dec. 20, 2024), <[Link] accessed 25
December 2025.
84
Centurion Indus. v. Warren Steurer and Assoc., 665 F.2d 323 (10th Cir. 1981).
85
MagicJack Vocaltec Ltd. v. [Link], Inc., No. 12-80360 (S.D. Fla. Oct. 18, 2012) (ECF No. 51).
86
Congoo, LLC v. Revcontent LLC, 2017 WL 3584205 (D.N.J. 2017).

Page | 21
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

production of source code would cause irreparable harm to their business. The Court
recognised the highly confidential nature, ruled that the plaintiff did not meet its
burden of proving that production of source code is relevant and necessary, and held
that it outweighs the need for production.87

60. It is humbly submitted that in the present case, the plaintiff has failed to prove that
the AI datasets used to train SwaRaj are necessary, and that its production is not
outweighed by the irreparable harm that would be caused to the Defendant. Reliance
is placed on the expert report which stated that the statistical similarity did not
establish source-based copying, and that the allegations of 82% similarity proving
copying was scientifically unsustainable in AI-generated media.88

61. Thus, it is argued, that compelling the disclosure of AI training data set of SwaRaj
is disproportionate and would irreparably harm the Defendants. It is also submitted
that the Plaintiff has not demonstrated that the examination of the entire training
dataset is indispensable.

4.3 Availability of alternatives defeats the need for compelled disclosure

62. It is submitted that if alternatives are available, then the source code need not be
compelled to be disclosed. In Telspace, LLC v. Coast to Coast Cellular, Inc., the
Court held that for disclosure to be warranted, the source code must be relevant and
necessary to the prosecution of the case, and if alternatives to disclosure are
available, a court would not be justified in ordering a disclosure.89

63. In Generac Power Sys. Inc. v. Kohler Co., the Court denied a motion to compel
source code because it was concerned about the code being used by the Plaintiff to
improve its products and believed that other substitutes or reasonable
accommodations were provided by the defendant including software demonstration
and manuals, which in the Court’s opinion would provide the same information.90

87
Ibid.
88
Moot prop, Annexure D.
89
Telspace, LLC v. Coast to Coast Cellular, Inc., 2014 WL 4364851, *4 (W.D. Wash. Sept. 3, 2014).
90
Generac Power Sys. Inc. v. Kohler Co., No. 11-1120 (E.D. Wisc. Jun. 6, 2012) (ECF No. 32).

Page | 22
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

64. In Transformative Learning Solutions Pvt. Ltd., v. Pawajot Kaur Baweja, the Delhi
HC held that in cases of patent or copyright infringement in source code of a
computer software, the need for a party to see the document may not arise as the
opinion with respect thereto is to be given by the expert only.91 Therefore, even if
the AI training data sets are relevant, the Defendant cannot be compelled to disclose
them, as the dispute regarding copyright infringement can be addressed via expert
report or limited technical inspection.

5. THE DEFENDANTS ARE NOT LIABLE TO PAY ROYALTY TO THE PLAINTIFF

65. It is submitted that the Plaintiff is not entitled to receive reasonable royalties. The
right is a statutory creation, not a common law right. Royalties are payable only
when a work is utilized in a manner specifically prescribed under Section 38A (e.g.,
reproduction, broadcast, or communication to the public).92

66. There exists no inalienable right to royalty because the payment of royalty is
contingent upon the commercial exploitation of the sound recording or the
recording of the performance itself which results in commercial benefits for the
licensee.93 The internal computational process of training an AI model does not fall
under such a category.

67. The process of training SwaRaj involves the conversion of audio data into
numerical vectors. The original recordings are never communicated to the public at
any stage of this process.94 As held in Myspace Inc., infringement requires the actual
publication or communication of the work to the public.95 Internal data processing
or training does not satisfy this threshold because the expressive elements of the
training data are not accessible by the users or communicated to them.

68. The Plaintiff may rely on Indian Singers Rights Association v. Night Fever Club but
such reliance is misplaced because the case involves materially different

91
Transformative Learning Solutions (P) Ltd. v. Pawajot Kaur Baweja, 2019 SCC OnLine Del 9229.
92
The Copyright Act 1957, s 38A.
93
The Copyright Act 1957, s 18(1).
94
Jenny Quang, ‘Does Training AI Violate Copyright Law’ (2021) 36 Berkey Technology Law Journal
<[Link] accessed 24 December 2025.
95
Myspace Inc. v. Super Cassettes Industries Ltd. 2016 SCC OnLine Del 6382.

Page | 23
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

circumstances. There, the defendants played songs verbatim in their premises to

entertain customers, enhance the ambience of their premises and derive commercial
gains.96 However, in the present case, the Defendants are using the non-expressive
elements of the data for only for computational analysis to create a new tool. The
SwaRaj app does not replay Shreya Sheeran’s songs, it generates new work.97

69. Tere Bin Main Kya does not reproduce the literary or musical expressions as used
in the Plaintiff’s musical works and performances in any manner. The similarity in
voice style is by virtue of training on a wide dataset comprising the techniques
routinely used by female South Asian singers.98 Further, the non-expressive use in
training qualifies the criteria for availing the Fair Dealing defence under Section 52.

70. In Authors Guild v. Google, Inc., the U.S. Second Circuit held that creating a digital
search database of millions of copyrighted books was transformative and fair use
because it provided a new service without substituting the original books.99
Similarly, the Defendants have analysed the unprotectable stylistic elements like
pitch, tempo and vibrato to train an algorithmic model.

71. As decided in R.G. Anand v. Delux Films, copyright protects only the expression,
not the underlying idea or technique.100 The AI learns the technique (how to sing
with vibrato), which is not copyrightable. Therefore, no royalty is due for learning
a skill, even if learned from a copyrighted source.

72. Even if the intermediate storage of data for training is considered a technical
reproduction, it is protected under Section 52(1)(a)(i) of the Copyright Act, which
permits fair dealing for ‘private or personal use, including research.’101 The
development of the V-23-F model by Ed Udhas was a research endeavour into deep

96
Indian Singers Rights Association v. Night Fever Club & Lounge 2016 SCC OnLine Del 5418.
97
Moot Proposition, ¶ 8 and 9.
98
Moot Proposition, Annexure F.
99
Authors Guild Inc. v. Google Inc., Case No. 13-4829-cv (2d Cir. 2015).
100
R.G. Anand v. Deluxe Films (1978) 4 SCC 118.
101
The Copyright Act 1957, s 52(1)(a).

Page | 24
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

neural networks.102 The subsequent commercialization of the tool does not

overweigh the transformative nature of the training phase.

73. Section 52(1)(w) specifically exempts the making of a three-dimensional object

from a two-dimensional artistic work for industrial application.103 This may be
extended by analogy of creating a functional AI tool from audio files.

74. It is submitted that the imposition of a royalty obligation for obtaining access to the
training data would exponentially increase the transaction costs involved in the
development of the Sindian AI industry due to the vast quantum of license fees
payable. The practical impossibility for determining the share of royalties for
performers of varying fame and the distribution of the same among all performers
whose works may further complicated the situation.104 Associated Electronic &
Electrical Industries v. Sharp Tools held that patent and copyright laws are meant
to encourage industry, not create a blockade against technological progress.105

75. The Plaintiff’s claim for "Reasonable Royalty" also fails the test of causation. The
V-23-F model is a blend of over one thousand vocal profiles. It is technically
impossible to determine how much value was derived specifically from the
Plaintiff’s tracks versus the 999 other singers. As argued in Issue I, there is also no
market substitution. The Defendants have not exploited the Plaintiff’s work in the
traditional sense and have only used the unprotectable non-expressive elements in
a transformative manner for training the AI. Therefore, they are not liable to pay
royalty or compensation to the Plaintiff.

6. THE PLAINTIFF IS NOT ENTITLED TO PERMANENT INJUNCTION RESTRAINING THE

DEFENDANTS FROM USING ANY VOICE OR VIDEO RESEMBLING HER VOICE OR

PERFORMANCES, OR USING ANY DATASETS CONTAINING HER PERFORMANCES

102
Moot Proposition, ¶ 7.
103
The Copyright Act 1957, s 52(1)(w).
104
Dorien Herremans, ‘Royalties in the Age of AI: Paying Artists for AI-Generated Songs’ (WIPO
Magazine, 6 May 2025) accessed <[Link]
the-age-of-ai-paying-artists-for-ai-generated-songs-73739> 28 December 2025.
105
Associated Electronics v. Sharp Tools AIR 1991 Kar 406.

Page | 25
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

76. It is humbly submitted that the Plaintiff is not entitled to any permanent injunction
restraining the Defendants. Firstly, the rights of the Plaintiff are still disputed. (6.1)
Secondly, a blanket restraint is overbroad, vague, and legally impermissible. (6.2)
Thirdly, no irreparable harm is established (6.3) and fourthly, a permanent
injunction would be detrimental to the economy. (6.4)

6.1 Plaintiff’s rights are not undisputed to be entitled to a permanent injunction

77. In Anathula Sudhakar v. P. Buchi Reddy, the SC held that a case for permanent
injunction lies only when the plaintiff’s right is clear and undisputed.106 It is
submitted that in the present case, the use of the Plaintiff’s voice, or performance
or any dataset containing the same is still under dispute and are yet to be tried.

78. The Defendant’s Expert Report undermines the Plaintiff’s case. While the Plaintiff
relies on the numeric similarity metrics, the counter-expert report expressly clarifies
that statistical similarity does not establish source-based copying in AI-generated
outputs, and such metrics are scientifically insufficient to conclude that SwaRaj was
trained on the specific voice of the Plaintiff.107

79. In F-Hoffmann-La Roche Ag v. Zydus Lifescience, the Delhi HC considering the

technical complexit, held that it may consider appointing an independent Scientific
Advisor for a neutral expert analysis, and to ensure a thorough exploration of the
technical issues, considered simultaneous questioning of experts from both sides.108

80. It is argued that the present case also deals with a technically complex matter, and
since two experts have given counter-expert reports, the Court ought to appoint an
independent expert for a neutral expert report, so that the Court can understand how
generative AI works in applications like that of SwaRaj. Until this is done, the
plaintiff cannot be said to be entitled to any permanent injunction

6.2 Blanket restraint is overbroad, vague and legally impermissible

106
Anathula Sudhakar v P. Buchi Reddy (2008) 4 SCC 594.
107
Moot Proposition, Annexure D.
108
F-Hoffmann-La Roche AG & Anr. v Zydus Lifesciences Ltd 2025 SCC OnLine Del 2347.

Page | 26
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

81. It is submitted that a grant of permanent injunction is impermissibly wide, as a

restraint against the use of any voice or video “resembling” that of the plaintiff lacks
objective standards and is incapable of precise enforcement.

82. In Shree Vardhman Rice and Gen Mills v. Amar Singh Chawalwala, the SC ordered
that matters relating to trademarks, copyrights, and patents should be finally
decided very expeditiously by the Trial Court instead of merely granting or refusing
to grant an injunction.109

83. Even in cases involving personality-rights, courts have restrained

misrepresentation, false endorsement, or direct exploitation, not mere stylistic
similarity. In Asha Bhosle v. Mayk Inc., the Plaintiff was a playback female singer
with a renowned reputation similar to Shreya Sheeran’s, and the Defendants had
utilised algorithms to offer the public at large the conversion of any voice or any
sound recording or song of their choice to the voice of the Plaintiff.110

84. In Arijit Singh v. Codible Ventures LLP, the Defendant enabled an AI tool to convert
any text, speech, voice recording or audio file to the Plaintiff’s AI voice version by
uploading a dataset comprising 456 songs from his repertoire.111

85. It is argued that, unlike these two cases, in the present case, the Defendant did not
directly convert any voice or sound recording to the voice of the Plaintiff, nor did
the Defendant commercially misuse the voice of the Plaintiff. The data sets of
SwaRaj are developed from publicly accessible audio recordings to “train”
generative models.112 There was never a case of the Defendants specifically using
the Plaintiff’s voice to generate songs and videos.

86. In Super Cassettes Industries v. Hamar Television Network, the Court cited Lord
Denning’s ruling in Hubbard & Anr v. Vosper and Anr., wherein it was held that the
right course is to look at the whole case, and that the judge must have regard to not
only the strength of the claim but also the strength of the defence and then decide

109
Shree Vardhaman Rice and General Mills v. Amar Singh Chawalwala (2009) 10 SCC 257.
110
Asha Bhosle v Mayc Inc 2025 SCC OnLine Bom 3485.
111
Arijit Singh v Codible Ventures LLP 2024 SCC OnLine Bom 2445.
112
Moot Proposition, ¶ 6.

Page | 27
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

what is to be done.113 Further, he also held that at times it is “best not to impose a
restraint upon the defendant” but to leave him free to go ahead.

87. It is argued, that the above was held in context of interim injunction, and in
pursuance of the same, the Defendants humbly request this Hon’ble Court to decide
the case on its merits before ordering for a grant of permanent injunction.

6.3 No irreparable harm is established

88. It is humbly submitted that the Plaintiff’s allegations of dilution, audience

confusion, and market erosion are unsupported by evidence, and remain
speculative.114 The plaintiff has placed no such material to show that the
Defendant’s actions have displaced her performances, contracts or goodwill.

89. Reliance is placed on Super Cassettes Industries v. Hamar Television Network,

wherein compact discs containing the infringement were filed as a document, and
based on that, the Delhi HC provided an injunctive relief.115 In the absence of such
proof, the restraint would be impermissible.

90. It is further submitted that awarding damages can sometimes also have the same
effect as granting an injunction, which in effect could push parties towards good-
faith negotiation, including counter-offers from the Defendant.116 Although this has
been the practice in Patent infringement cases, the core principle can be applied in
the present case as well.

6.4 Grant of permanent injunction would be detrimental to the economy

91. It is submitted that generative AI has huge potential to benefit the global economy.
It has been held that Generative AI’s impact could “add trillions of dollars in value

113
Super Cassettes Industries Ltd v Hamar Television Network (P) Ltd 2010 SCC OnLine Del 2086.
114
Moot Proposition, ¶ 26.
115
Super Cassettes Industries Ltd v Hamar Television Network (P) Ltd 2010 SCC OnLine Del 2086.
116
Yogesh Pai, ‘Patent Injunction Heuristics in India’ (2019) KLUWER LAW INTERNATIONAL BV
<[Link] accessed 25 December 2025.

Page | 28
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

to the global economy”.117 As per UNCTAD’s global survey, creative economy

employs “between 0.5% to 12.5% of the workforce” across countries118.

92. For Sindia to stay competitive with other countries, it is important to let applications
such as SwaRaj innovate and grow, while also ensuring fair rules and protection of
artists and writers. It is an issue of economic priority, and there is significant legal
uncertainty affecting both technology innovators and content creators.119 A blanket
ban would prevent Sindia from staying ahead in the race of digital world.

93. It is further submitted, that countries across the world are dealing with extending
copyright protection to AI generated content. For example, EU and Japan emphasise
the need for “significant human input” standard for AI generated content.120 India
is also currently developing AI copyright policy. In such situation, a permanent
injunction on the Defendants, would go against the interests of Sindia.

94. Therefore, it is humbly submitted that there is a need for a balanced framework that
enables AI innovation while also respecting creators’ rights, and if a permanent
injunction is granted by this Hon’ble Court, then it would stifle all kinds of AI
innovation and harm overall public benefit.

117
McKinsey & Company, ‘The economic potential of generative AI: The next productivity
frontier’,(McKinsey & Company, 14 June 2023)
<[Link]
ights/the
%20economic%20potential%20of%20generative%20ai%20the%20next%20productivity%20frontier/th
[Link]> accessed on 28 December
2025.
118
UNCTAD, ‘Creative Economy Outlook 2024’ (UNCTAD, 11 July 2024) accessed on 29 December
2025.
119
Ms. Astha Ojha, ‘AI & Copyright in India: Law, Policy, and the Future of Creative Rights’ (NeGD, 8
October 2025) <[Link]
[Link]> accessed on 29 December 2025.
120
Ibid.

Page | 29
MEMORIAL for DEFENDANTS
 1ST CNLU-DPIIT IPR NATIONAL MOOT COURT COMPETITION, 2026

PRAYER

Wherefore, in the light of the issue raised, arguments advanced, and authorities cited,
the Defendants pray that this Hon’ble Court may be pleased to adjudge and declare:

I. The Plaintiff is not entitled to an interim injunction restraining the Defendants

from using any voice/video resembling her voice/performance pending disposal
of the suit

II. The use of AI by the Defendants for training a voice model, and generating the
song “Tere Bin Main Kya” does not amount to an infringement under the
Copyright Act, 1957.

III. The alleged replication of the plaintiff’s vocal identity and performance style
does not violate her personality rights under Sindian law

IV. The Defendants cannot be compelled to disclose their AI training datasets and
model documentation

V. The Defendants are not liable to pay royalty or compensation to the Plaintiff

VI. The Plaintiff is not entitled to any permanent injunction restraining the
Defendants

AND/OR

Any other order, direction, or relief that this Hon’ble Court may deem fit in the
interests of justice, equity, expediency, and good conscience.

For this Act of kindness, the respondents, as is duty bound, shall ever pray.

All of which is respectfully submitted,

On behalf of the Defendants

COUNSELS for DEFENDANTS

Page | 30
MEMORIAL for DEFENDANTS