Md. Yasin
HILSR, Jamia Hamdard
Liba Fatima
HILSR, Jamia Hamdard
Md. Yasin
HILSR, Jamia Hamdard
Liba Fatima
HILSR, Jamia Hamdard
Deepfakes and due process: reassessing evidentiary integrity under india’s new criminal codes
I. Introduction
When India’s new criminal codes, the Bharatiya Nyaya Sanhita 2023 (BNS), Bharatiya Nagarik Suraksha Sanhita 2023 (BNSS), and Bharatiya Sakshya Adhiniyam 2023 (BSA), came into force in July 2024, they marked the end of a colonial legal era. Yet this moment of legislative renewal arrived alongside an unprecedented technological disruption i.e the rise of “deepfakes”. These AI-generated images, voices, and videos now blur the boundary between truth and illusion, striking at the very heart of evidentiary trust on which justice depends.
While the BSA modernises terminology, it remains technologically agnostic. This article argues that procedural reform not merely punitive legislation is essential to preserve due process and evidentiary integrity under Article 21 of the Constitution. It analyses doctrinal gaps in the BSA, reviews recent judicial practice, and proposes a calibrated “Section 63A” amendment to anchor synthetic media authentication within India’s new evidentiary regime.
II. EVIDENTIARY INTEGRITY IN THE AGE OF DEEPFAKES
The credibility of evidence rests on three pillars i.e Authenticity, Reliability and Admissibility.1 Deepfakes shake each of these foundations. They can be manufactured without any originating device, rendering source verification nearly impossible. They erode reliability by allowing imperceptible manipulation and they blur admissibility itself, leaving courts struggling to tell fact from fabrication.
Under Section 63 BSA2, electronic records remain admissible only if accompanied by a certificate of authenticity. Yet that assumption presumes human-generated data. Deepfakes being algorithmic composites, may have no certifiable origin.
Section 39(2) BSA3 empowers courts to seek expert opinion on electronic evidence. But as of mid 2024, India’s forensic capacity is embryonic. The National Forensic Science Infrastructure Enhancement Scheme 2024 (₹ 2,254 crore) aims to establish AI-enabled forensic units in every State by 2027, but most trial courts presently lack access to certified laboratories.4 Without such capacity, procedural compliance risks becoming symbolic.
III. STATUTORY LANDSCAPE
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules 20235 imposed India’s first statutory obligation on platforms to remove impersonating or deepfake content within 36 hours.⁵ Rule 3(1)(b)(v) prohibits “impersonation of another person.”
Subsequently, the IT (Amendment) Rules 2025,6 strengthened this duty by mandating
watermarking of AI-generated content and 24-hour takedown. These advances, though valuable
for platform governance, remain detached from courtroom authentication. No Indian statute yet explains how synthetic content should be proved once tendered in evidence.
Meanwhile, Section 67 BNS 2023 criminalises the publication of obscene or false electronic material but provides no evidentiary procedure for establishing falsity. Law thus condemns the act but not the method of proof.
IV. JUDICIAL ENGAGEMENT (2023–2025)
India’s higher judiciary has begun acknowledging the evidentiary hazards posed by synthetic media. The most instructive cases are as follows:
Case | Court & Year | Evidentiary Relevance |
Anil Kapoor v. Simply Life India7 | Delhi HC, 2023 | Acknowledges deepfake impersonation and recognized the Synthetic media harm. |
Amitabh Bachchan v. Unknown8 | Delhi HC, 2024 | AI voice cloning recognised as a violation of personality rights |
Asha Bhosle v. Mayk Inc & Ors9 |
Bombay HC, Oct 2025 | This order extends the evidentiary horizon by explicitly requiring preservation of digital logs, a precedent for admissibility and forensic verification under the forthcoming IT Rules. |
These orders reflect a growing judicial maturity, shifting the focus from protecting reputation to ensuring procedural authenticity in how digital evidence is treated. They affirm that personality rights and evidentiary safeguards are now converging terrains of digital justice.
V. COMPARATIVE FRAMEWORKS AND INDIAN FEASIBILITY
As held in Vishaka v. State of Rajasthan10 and reiterated in People’s Union for Civil Liberties v. Union of India,11 Indian courts borrow from international and comparative law only when doctrinally compatible with constitutional principles. This contextual approach was reaffirmed in Justice K.S. Puttaswamy Case12 where the Supreme Court endorsed comparative reference as an interpretive aid—but cautioned that the ultimate contours of rights must remain anchored in India’s constitutional text and ethos. Three systems illustrate boundaries of adaptation i.e:
- European Union: The Artificial Intelligence Act 2024 (Reg. 2024/1689, (Arts. 50 and 53),13 requires that AI-generated content be transparently disclosed and, where possible, watermarked to ensure traceability. Although the Act’s supranational enforcement mechanisms make direct transplantation to India impractical, a parallel model could be adopted domestically through a “traceability certificate” framework under Section 79A of the Information Technology Act 2000.
- United States: Rule 901(a) of the Federal Rules of Evidence14 requires that evidence be supported by proof that it “is what the proponent claims it is.” In American courts, this often means demonstrating authenticity through metadata or blockchain hash verification. India may not yet have the same digital infrastructure, but the underlying idea holds, whoever seeks to rely on an electronic record must also prove that it is genuine, a burden that naturally falls under Section 107 of the Bharatiya Sakshya Adhiniyam, 2023.
- United Kingdom: Section 69 of the Police and Criminal Evidence Act 198415 required proof that a computer was functioning properly at the time a record was produced. Although later repealed, its evidentiary rationale verifying the operational integrity of digital systems remains persuasive. India could adapt this principle through expert certification under Section 39(2) of the Bharatiya Sakshya Adhiniyam, 2023, using laboratories notified under Section 79A of the Information Technology Act, 2000.
The comparative lesson is crystal clear that India should imitate process, not form. Foreign evidentiary models work because they stand on mature forensic ecosystems. Before importing their procedures, India must first invest in building the institutional and technical capacity that makes authenticity provable, not presumed.
VI. JUDICIAL TRENDS AND REFORM PATHWAYS
India’s approach to deepfakes reflects both progress and unease. Courts have shown commendable sensitivity to the reputational harm that synthetic content can inflict, yet they continue to struggle with the procedural question of how such material should be tested and proved. Deepfakes test the very constitution of justice, eroding truth, authenticity, and trust. The Bharatiya Sakshya Adhiniyam, 2023 refines form without substance, and the Information Technology Rules, 2025 police content more than proof. Real reform demands clear law, agile procedure, and forensic strength—lest illusion eclipse fact and fairness in India’s courts.
(a) Proposed Insertion of Section 63A, Bharatiya Sakshya Adhiniyam, 2023
“ Section 63A: Synthetic Media Evidence
- i. Where any evidence is suspected to be synthetic media or algorithmically generated, the Court shall direct forensic examination by a laboratory notified under Section 79A of the Information Technology Act, 2000.
- ii. Non-disclosure of the use of generative artificial intelligence shall invite an adverse inference under Section 119, Illustration (g) of the Bharatiya Sakshya Adhiniyam, 2023.
- iii. The Central Government may, by notification, prescribe the standards and protocols for forensic verification of synthetic media, including watermarking, metadata preservation, and authenticity certification.”
This formulation fits within Parliament’s competence under Entry 12, List III, and enforces procedural fairness under Article 21.16
(b) Procedural Integration:
under Section 349(4) BNSS, 2023 empowers courts to issue electronic summons or production orders. A practice direction could require that any party tendering digital media file, an accompanying Deepfake Disclosure Affidavit confirming whether AI tools were used. This embeds verification within existing procedural powers without statutory overhaul.
(c) Institutional Capacity:
Under the NFSI 2024 Scheme,17 each State will house an AI- enabled forensic division by 2027. Courts should be authorised to requisition these units via notifications u/s 39(2) BSA. Judicial academies must include AI-forensics training to translate technical reports into admissible reasoning.
(d) Constitutional Dimensions:
The right to a fair trial under Article 21 demands procedures that are just, fair and reasonable.18 Conviction or defamation based on unverified synthetic evidence would offend this guarantee. At the same time, Articles 245 and 24619 mark the outer limits of legislative competence. Any reform of evidentiary procedure must come through Parliament, not through judicial improvisation, so that the balance between innovation and separation of powers reaffirmed in Article 141 remains
VII. CONCLUSION
Deepfakes represent more than a technological upheaval they strike at the constitutional heart of evidentiary justice. The Bharatiya Sakshya Adhiniyam, 2023 updates the vocabulary of proof but not its substance, while the platform duties introduced under the 2025 IT Rules may curb online harm yet do little to secure the authenticity of evidence inside a courtroom.
A credible national response must move in step with technology on three fronts. Parliament should begin by amending the Bharatiya Sakshya Adhiniyam, 2023 to insert Section 63A, explicitly recognising and regulating the authentication of synthetic media. Procedural law must follow suit through Deepfake Disclosure Affidavits under Section 349(4) of the Bharatiya Nagarik Suraksha Sanhita, 2023, and Order XI Rule 14 of the Code of Civil Procedure, requiring parties to disclose any AI-generated evidence before trial. Institutional capacity must also keep pace with these reforms, the National Forensic Science Infrastructure Scheme, 2024 must be implemented in full to establish AI enabled forensic units across States and to train judges to evaluate such evidence with skill and discernment.
Until Parliament acts, the judiciary should invoke Section 349(4) of the BNSS creatively, issuing practice directions that make AI-disclosure affidavits mandatory. Delay threatens more than procedural confusion; it endangers the credibility of justice itself, allowing algorithmic illusion to eclipse verifiable truth.