Authored By: Sonal Didwania
ICFAI LAW SCHOOL, IFHE, HYDERABAD
India replaced its colonial evidence statute in 2023, and with it came a promise: that the law governing proof in Indian courtrooms would finally speak the language of the digital age. The Bharatiya Sakshya Adhiniyam,2023, brought electronic records within the fold of documentary evidence, updated certification requirements for computer-generated material, and introduced a hash-value mechanism meant to guarantee that a record has not been altered after it was created. What it did not do, and what its drafters could not have fully anticipated in the years leading up to its passage, was reckon with a technology capable of manufacturing a video or an audio clip that never corresponds to any real event at all. That technology, generative synthetic media commonly called deepfakes, now sits at the center of a widening gap between what Indian evidence law verifies and what it actually needs to verify.
This article argues that the gap is structural rather than incidental. Section 63 of the BSA, like Section 65B of the statute it replaced, exists to answer one question: was this record produced by a properly functioning device, and has it stayed unaltered since it was made? That question is about provenance, not truth. Deepfake technology does not corrupt a record after it is created; it fabricates the record at the moment of creation. A synthetic video can pass every procedural safeguard the BSA demands and still show a conversation, a confession, or an event that never actually happened.
The Statutory Architecture
The BSA, which came into force on 1 July 2024, treats electronic records as documents for admissibility purposes.[1] Sections 61 and 62 place electronic and paper records on an equal evidentiary footing, while Section 63 sets out the conditions on which an electronic record becomes admissible without the maker being called to testify in person.[2] Section 63(4) requires a certificate identifying the device that produced the record, describing how the record was produced, and, notably, specifying a hash value that can later verify whether the record has been altered.[3] The certificate is now split into two parts: one completed by the party tendering the evidence and a second completed by a qualified expert.[4] This structure closely tracks the framework that governed Section 65B of the Indian Evidence Act, 1872, as interpreted by the Supreme Court in Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal, which held that such certification is mandatory before secondary electronic evidence can be admitted.[5]
None of these provisions ask whether the underlying audio, video, or image was genuine at the moment of capture. They ask whether the chain of custody after capture has been documented and whether the device functioned as claimed. A hash value confirms that a file has not been tampered with since it was hashed; it says nothing about whether the file was synthetic from the outset.[6] The Information Technology Act, 2000 supplies the surrounding criminal-law framework: Section 66C punishes identity theft through the fraudulent use of another person’s electronic signature or password, Section 66D punishes cheating by personation through a computer resource, and Section 66E punishes the capture or publication of images of a person’s private areas without consent, while Section 79 sets out the corresponding intermediary liability rules.[7] That statute predates commercially accessible generative video tools by nearly two decades and contains no provision naming synthetic or AI-manipulated media as a distinct category of harm.
Judicial Engagement So Far
Indian courts have begun to confront synthetic media, though largely at the margins of personality-rights and privacy law rather than through the evidentiary provisions directly. In Anil Kapoor v. Simply Life India & Ors., the Delhi High Court granted an ex parte, omnibus injunction restraining sixteen defendants and, in effect, the world at large, from using the actor’s name, voice, image, or persona through artificial intelligence tools including deepfakes and face morphing.[8] Justice Prathiba M. Singh grounded the order in the constitutional right to privacy under Article 21 and in the common-law tort of passing off, drawing on both R. Rajagopal v. State of Tamil Nadu and comparative authority such as the American case of White v. Samsung Electronics America.[9] The judgment is a significant marker of judicial willingness to act against synthetic-media misuse, but it addresses reputational and dignitary harm to a named plaintiff; it does not, and was not designed to, resolve how a court should treat a deepfake tendered as evidence of a disputed fact in an ordinary civil or criminal trial.
On the evidentiary side proper, courts have so far repeated the general caution that has long accompanied electronic evidence. In Shafhi Mohammad v. State of Himachal Pradesh, the Supreme Court, before being overruled on the narrow question of mandatory certification in Arjun Panditrao itself, nonetheless underscored that electronic evidence remains susceptible to tampering and manipulation and that courts must apply care in relying on it, a caution the Court did not disturb even as it tightened the certification rule.[10] That caution is doctrinally sound but practically thin: a trial judge told to scrutinise authenticity has no forensic standard, no accredited laboratory protocol, and no statutory definition of deepfake or synthetic media to scrutinise against. The instinct is right; the tools to act on it are missing.
III. Locating the Gap
The gap can be stated precisely. Evidence law traditionally polices two distinct risks: the risk that a genuine record has been altered after creation, and the risk that a witness or document is lying about what it purports to show. The BSA’s certification regime is finely tuned to the first risk. Deepfake technology introduces a third risk that fits neither category: a procedurally pristine recordWriting, never altered after capture because there was no authentic capture to alter, and yet substantively false from its first frame.[11]
This has at least three practical consequences. First, it inverts the burden of proof in ways the drafters likely did not intend. Once a party produces a properly certified recording, the opposing party is functionally required to prove a negative, that the event depicted did not occur, rather than merely raising doubt about the record’s provenance.[12] Second, it creates an incentive for courts to overcorrect by treating all electronic evidence with generalised suspicion, which risks undermining the very digitisation of evidence that the BSA was meant to encourage.[13] Third, and most concretely, India currently lacks any government-accredited forensic laboratory or standardised judicial protocol for authenticating suspected synthetic media, meaning that even a judge alert to the problem has no institutional mechanism to resolve it.[14] The Law Commission of India has not yet produced a report specifically addressing deepfake forensics in evidence law, leaving the field without even the customary first step toward statutory reform.[15]
The 2026 Intermediary Rules: A Partial Answer
The most recent regulatory movement comes not from evidence law but from platform regulation. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 introduce obligations on significant social media intermediaries to label, disclose, and expeditiously act upon synthetically generated information, including compressed takedown timelines once such content is flagged.[16] These rules are useful for containing the circulation of synthetic media online, but they operate upstream of the courtroom. A labelling requirement imposed on a platform does nothing to help a trial judge who is handed a video, already downloaded, already stripped of platform metadata, and asked to rule on its admissibility and weight. Regulation of circulation and regulation of proof are different problems, and the 2026 Rules solve only the first.
Toward a Statutory Response
Closing the gap does not require abandoning the BSA’s certification model, which remains sound for its intended purpose. It requires supplementing that model with an authenticity layer specific to synthetic media. Three reforms would move matters forward considerably.
First, Section 63 could be amended to require, for audio-visual evidence tendered in contested proceedings, an additional forensic declaration addressing the likelihood of synthetic generation, separate from the existing device-and-chain-of-custody certificate. Second, a designated forensic authority, whether housed within the existing Central Forensic Science Laboratories or created afresh, should be empowered to issue authoritative findings on suspected synthetic media that courts can rely upon as a matter of course, rather than leaving each trial court to improvise scrutiny on an ad hoc basis. Third, the Bharatiya Nyaya Sanhita’s provisions on fabricating false evidence, drafted with human forgers rather than generative models in mind, should be extended to expressly cover the creation or knowing use of synthetic media intended to deceive a court, closing the gap the Jus Corpus commentary has identified between existing fabrication offences and AI-generated fabrication.[17]
Comparative practice offers some direction without offering a template ready for transplant. The European Union’s AI Act imposes transparency obligations requiring the labelling of AI-generated or manipulated content, a regulatory posture similar in spirit to India’s 2026 Intermediary Rules and subject to the same limitation of operating outside the courtroom.[18] No major jurisdiction has yet produced a fully worked evidentiary framework purpose-built for adjudicating deepfake authenticity, which means India’s legislature and judiciary are, on this narrow question, working at the frontier rather than catching up to settled international practice.
Conclusion
The Bharatiya Sakshya Adhiniyam, 2023 modernised Indian evidence law for the era of the computer record, but it was drafted for a world in which the central evidentiary anxiety was tampering after the fact. Generative synthetic media has introduced an anxiety of a different order, fabrication at the point of origin, and the statute’s certification architecture, however carefully constructed, was not built to detect it. Judicial caution in cases such as Shafhi Mohammad signals awareness of the underlying risk that electronic records can mislead; the injunction in Anil Kapoor signals a willingness to act decisively where personality rights are threatened. Neither response, however, substitutes for a dedicated statutory and institutional mechanism capable of testing whether a recording placed before a court ever corresponded to reality at all. Until that mechanism exists, Indian courts will continue to ask electronic evidence the right questions about custody and continuity while leaving unasked the one question that now matters most.
Reference(S):
[1]Bharatiya Sakshya Adhiniyam, No. 47 of 2023, ss. 61-62 (India).
[2]Bharatiya Sakshya Adhiniyam, No. 47 of 2023, s. 63 (India).
[3]Bharatiya Sakshya Adhiniyam, No. 47 of 2023, s. 63(4)(c) (India).
[4]Id.
[5]Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal, (2020) 7 S.C.C. 1 (India).
[6]See Deepfakes & Electronic Evidence: Can Courts Still Trust What They See & Hear?, Tax Guru (2026), https://taxguru.in/corporate-law/deepfakes-electronic-evidence-courts-trust-hear.html.
[7]Information Technology Act, No. 21 of 2000, ss. 66C, 66D, 66E, 79 (India).
[8]Anil Kapoor v. Simply Life India & Ors., CS(COMM) 652/2023 (Del. H.C. Sept. 20, 2023) (India).
[9]R. Rajagopal v. State of T.N., (1994) 6 S.C.C. 632 (India); White v. Samsung Elecs. Am., Inc., 971 F.2d 1395 (9th Cir. 1992).
[10]Shafhi Mohammad v. State of Himachal Pradesh, (2018) 2 S.C.C. 801 (India), overruled in part on the certification requirement by Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal, (2020) 7 S.C.C. 1 (India).
[11]See Artificial Reality vs. Judicial Truth: Deepfake Evidence in Indian Courts, Jus Corpus (2026), https://www.juscorpus.com/artificial-reality-vs-judicial-truth/.
[12]See Deepfake Proof in Indian Criminal Trials: A Challenge to Indian Courts, Jus Corpus (2025), https://www.juscorpus.com/deepfake-proof-in-indian-criminal-trials-a-challenge-to-indian-courts/.
[13]Id.
[14]See Artificial Reality vs. Judicial Truth, supra note 11.
[15]Id.(noting the absence of Law Commisssion report on deepfake forensics)
[16]Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 (India).
[17]Bharatiya Nyaya Sanhita, No. 45 of 2023, s. 192 (India); see Artificial Reality vs. Judicial Truth, supra note 11.
[18]Regulation 2024/1689, of the European Parliament and of the Council of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence, art. 50, 2024 O.J. (L 1689) (EU).
