SEEING IS NO LONGER BELIEVING: DEEPFAKE EVIDENCE AND THE BHARATIYA SAKSHYA ADHINIYAM, 2023

Artificial intelligence has transformed the evidentiary assumption that photographs, videos and voice recordings provide comparatively objective representations of reality. Deepfakes - synthetic or materially manipulated audio-visual content created through artificial intelligence and machine-learning techniques can convincingly attribute words, conduct and events to individuals who never spoke, acted or appeared as represented. Their emergence therefore creates a dual evidentiary danger: fabricated media may be accepted as genuine, while authentic recordings may be dismissed through an allegation that they are AI-generated, a phenomenon often described as the “liar’s dividend.” The Bharatiya Sakshya Adhiniyam, 2023 (BSA), which came into force on 1 July 2024, modernises Indian evidence law by expressly recognising electronic and digital records, expanding the conception of primary evidence and establishing a detailed framework under Section 63 for computer outputs. However, a crucial distinction remains between admissibility, integrity and authenticity. A Section 63 certificate or cryptographic hash may demonstrate the manner in which a digital file was produced and whether the submitted copy remained unchanged, but neither necessarily proves that the depicted event actually occurred or that the voice, face or conduct represented in the file belongs to the apparent individual. This paper examines the treatment of deepfake evidence through Sections 2, 39, 57, 61 - 63, 86 and related provisions of the BSA, traces the judicial evolution of electronic evidence authentication, and analyses recent Indian developments concerning synthetic media, personality rights and AI-generated content. Particular attention is given to the Information Technology Amendment Rules introduced in February 2026, which establish duties concerning synthetically generated information, labelling and provenance. The paper argues that India requires a provenance-oriented and technologically neutral evidentiary framework combining chain of custody, native-file preservation, metadata analysis, cryptographic verification, forensic expertise, platform records and independent corroboration. Such an approach can preserve the probative potential of digital evidence without permitting either technological deception or unsupported allegations of manipulation to undermine the administration of justice.