INTRODUCTION
The Bharatiya Sakshya Adhiniyam (BSA) 2023, a new law in India, expressly permits the use of electronic records as evidence in place of the previous colonial Evidence Act 1872. Under Sections 61–63 of the BSA,[1] courts admit electronic data as evidence. Deepfakes, which can produce incredibly realistic artificial picture and sound recordings that replicate an individual’s activities and conversations, are the result of recent advances in artificial intelligence.
A new problem arises from such creations: although it could be feasible to certify such material as admissible evidence, the material itself would be completely fake and hence unlawful. According to legal experts, AI-generated manipulation undermines the credibility and reliability of audiovisual evidence. Courts may mistakenly accept fabricated media or question authentic evidence, thereby undermining the fairness of judicial proceedings. This blog argues that while the BSA 2023 modernises the admissibility of electronic evidence, it remains inadequately equipped to address the evidentiary risks posed by sophisticated AI-generated deepfakes.
DEEPFAKES AND THE EMERGING THREAT TO ELECTRONIC EVIDENCE
Deepfakes are AI-generated or AI-manipulated forms of media that appear authentic despite depicting events that never occurred. They may take the form of manipulated videos, fabricated images, voice cloning, or entirely synthetic audiovisual content capable of creating false impressions of real events.
Deepfakes undermine the reliability of audiovisual evidence by making fabricated content appear authentic. They may be used to create false confessions, fabricated crime scenes, or misleading recordings that distort judicial fact-finding. Deepfakes can be highly realistic, making fabricated content difficult to distinguish from authentic recordings. This creates particular challenges for courts that increasingly rely on digital photographs, videos, and audio recordings to establish disputed facts.
ADMISSIBILITY AND RELIABILITY OF ELECTRONIC EVIDENCE IN INDIA
Indian law recognises electronic records as admissible evidence subject to authentication requirements. The Bharatiya Sakshya Adhiniyam, 2023 largely retains the framework previously contained in Section 65B of the IEA[2] by requiring certification and proof of authenticity before electronic records may be relied upon in court.
In Arjun Panditrao Khotkar v Gorantyal,[3] the Supreme Court emphasised the importance of authenticity and procedural safeguards in relation to electronic evidence. This distinction becomes critical in the context of deepfakes, where authenticity of origin does not guarantee factual accuracy.
Even with the aforementioned formal procedures for determining admissibility, the authentication procedure is still required to demonstrate the legitimacy of recognized digital evidence. However, authentication alone cannot establish the factual accuracy of digital content. According to the legislation, an electronic document must be authentic if it was produced by a specific computer or camera and has not been altered.[4]
LEGAL CHALLENGES POSED BY DEEPFAKE EVIDENCE
- Verification Challenge: Because deepfakes are designed to appear authentic, traditional forensic indicators such as timestamps and metadata may be insufficient to reliably identify sophisticated deepfakes.
- Burden of Proof on the Opposing Party: Once Bharatiya Sakshya Adhiniyam’s legal requirements are met, it is presumed that the electronic proof is genuine. As a result, the burden shifts to the opposing party to demonstrate that the information is false.
- Lack of Expertise in Artificial Intelligence and Digital Forensic Methods: Addressing deepfakes requires technical expertise in AI and digital forensics. Many judges and lawyers lack specialised technical training.[5]
- Risk of Wrongful Conviction and Miscarriage of Justice: False photos, videos, or audio snippets may be used to wrongly implicate someone and result in their wrongful conviction. In Mendones v Cushman & Wakefield, Inc.,[6] a suspected deepfake was submitted as authentic evidence before concerns were raised regarding its authenticity, illustrating the growing difficulty of distinguishing genuine recordings from AI-generated fabrications.
- Constitutional Issues: The deepfakes may infringe upon the rights to privacy and dignity guaranteed by Article 21 of the Constitution.[7] By furnishing false evidence, they can even violate the right to a fair trial. Legal scholars have identified a phenomenon known as the “Liar’s Dividend”. As deepfakes become more common, individuals may challenge the authenticity of genuine evidence by claiming that it has been manipulated through artificial intelligence. This may increase evidentiary disputes and weaken confidence in the judicial process.[8]
IS INDIA’S EXISTING EVIDENCE FRAMEWORK ADEQUATE?
The BSA is a positive development because it formally recognises digital records and provides greater legal certainty. Deepfake technology, however, exploits the new law’s gaps. For instance, Section 63(4) of the BSA[9] describes a certificate format designed for conventional electronic data. The provision assumes the certifying officer knows how the data was generated. This creates a significant evidentiary gap because courts may be asked to evaluate digital material without knowing whether artificial intelligence played a role in its creation.
Experts highlight this. “Although the BSA has introduced several reforms, the court will need to authenticate each individual AI-altered image or video in order to negate any semblance of trial efficiency,” according to an analysis done by IJNTI.[10] According to an additional study, “the BSA 2023 cannot be said to be effective in merely certifying digital data unless it is coupled with efficient forensic techniques,” despite the fact that watermarks and AI technologies can identify deepfakes. This suggests that existing regulations alone may be inadequate to address deepfakes.
In India, there is currently no legislation specifically addressing deepfakes. The Information Technology Act (e.g., Sections 66E/67[11] deal with the privacy and obscenity of digital content) or standard criminal laws (e.g., fraud or defamation) may be used to punish any damaging AI content. But there doesn’t appear to be any clear guidelines for proof produced by AI. Courts must therefore force-fit such issues into existing legal frameworks. Similar concerns have emerged internationally. In Breton v Ministry of Health and Social Services,[12] the Commission d’accès à l’information du Québec acknowledged that AI-generated evidence can mislead courts and stakeholders, thereby compromising the integrity of the judicial process. Despite these concerns, the Commission admitted the evidence based on corroborating materials and probative value. The decision highlights challenges that Indian courts may increasingly face.
THE WAY FORWARD: REIMAGINING ELECTRONIC EVIDENCE IN THE AI ERA
Certain rules pertaining to AI evidence were suggested by the Law Commission of India in 2024. The proposals recommended specialised certification requirements for AI-generated and deepfake evidence and called for the express recognition of deepfakes in legislation. They further suggested that multimedia evidence suspected of being AI-generated should be authenticated by authorised forensic laboratories before admission. India may also consider stricter authenticity requirements and mandatory disclosure of AI-generated content used as evidence. Deepfakes differ from conventional electronic records because the challenge is not merely proving the source of the evidence but determining whether the content itself reflects reality. Consequently, evidentiary standards for AI-generated material may need to be more rigorous than those applied to ordinary digital records.[13]
In addition, judicial training, specialised forensic capabilities, AI verification laboratories, and clear guidelines for handling AI-generated evidence should be developed. India may also draw upon emerging international approaches, including the EU AI Act and evolving US standards on electronic evidence.
Courts should develop guidelines requiring disclosure of the source, editing history, and AI tools used to create disputed audiovisual evidence. Such measures would improve transparency and assist judicial assessment.
CONCLUSION
The challenges that electronic evidence faces in the age of artificial intelligence are starkly highlighted by the advent of deepfake technology. While India’s evidence laws have undergone significant modernization due to the Bharatiya Sakshya Adhiniyam, 2023, the rules pertaining to the admissibility of digital evidence focus more on the admissibility and authentication of electronic evidence than on the use of artificial intelligence to detect fraud. As artificial intelligence advances, the conventional methods of electronic document verification may become inadequate. Therefore, stronger forensic methods and clearer legal standards for AI-generated evidence are required. However, the issue is not whether or not courts should permit electronic evidence, but rather whether or not they can distinguish between authentic and fraudulent electronic data.
Author(s) Name: Gungun Sharma (Prestige Institute of Management and Research, School of Law, Deemed to be University Indore)
References:
[1] Bharatiya Sakshya Adhiniyam 2023, ss 61–63
[2] Indian Evidence Act 1872, s 65B
[3] Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal and Ors (2020) 7 SCC 1
[4] Anne Merminod et al., ‘Authenticating Evidence in the Age of AI’ (International Bar Association, 14 April 2026) <https://www.ibanet.org/authenticating-evidence-in-the-age-of-AI > accessed 24 June 2026
[5] Jared Perlo, ‘AI-generated evidence is showing up in court. Judges say they’re not ready.’ (NBC News, 18 November 2025) <https://www.nbcnews.com/tech/tech-news/ai-generated-evidence-deepfake-use-law-judges-object-rcna23597> accessed 24 June 2026
[6] Mendones v Cushman & Wakefield, Inc. [2025] Case No 23CV028772
[7] Constitution of India 1950, art 21
[8] Connor Heaton, ‘AI-Generated Evidence Is a Threat to Public Trust in the Courts’ (National Center for State Courts, 24 February 2026) <https://www.ncsc.org/resources-courts/ai-generated-evidence-threat-public-trust-courts> accessed 24 June 2026
[9] Bharatiya Sakshya Adhiniyam 2023, s 63(4)
[10] Pinky Bansal, ‘FUTURE CHALLENGES OF ELECTRONIC EVIDENCE: DEEPFAKES, AI DATA, AND BLOCKCHAIN UNDER BSA, 2023’ (2025) 3(10) International Journal of Novel Trends and Innovation <https://rjpn.org/ijnti/papers/IJNTI2510008.pdf> accessed 24 June 2026
[11] Information Technology Act 2000, ss 66E, 67
[12] Breton v Ministry of Health and Social Services [2025] QCCAI 280
[13] Merminod (n 4)