India's BSA: AI-Generated Evidence Sakshya Adhiniyam Authentication Framework

• The Bharatiya Sakshya Adhiniyam, 2023, does not categorically bar AI-generated documentary evidence, treating it similarly to other electronic records.
• Section 61 of the BSA grants electronic records the same legal validity as paper documents, without distinguishing between conventional software and generative AI outputs.
• Supreme Court rulings in *Anvar P.V. v. P.K. Basheer* and *Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal* established the mandatory nature of the Section 63 BSA certificate for electronic evidence, setting a procedural framework.
• The core legal challenge for AI-generated evidence has moved from admissibility to rigorous authentication, requiring proof of provenance, reliability, and independent corroboration.
• Generative AI poses unique authentication challenges because it can create content identical to authentic documents without a physical counterpart, unlike traditional electronic or photocopied evidence.

For AI-generated evidence, the primary legal hurdle has decisively shifted from mere admissibility to the rigorous demands of authentication.

The Bharatiya Sakshya Adhiniyam, 2023 (BSA) establishes a legal framework that does not impose any categorical prohibition on the use of AI-generated documentary evidence. The statute defines a "document" broadly, encompassing any output capable of conveying information through various means like letters, figures, or marks, irrespective of the method used for its creation. This expansive definition ensures that materials produced by artificial intelligence systems fall within the scope of admissible evidence.

Furthermore, Section 61 of the BSA explicitly grants electronic or digital records the same legal effect, validity, and enforceability as traditional paper records, without any qualification. This legislative stance indicates that the statutory scheme does not differentiate between documentary evidence generated by conventional software and that produced using generative AI. The primary concern of the law is the legal status and proof of electronic records themselves, rather than the specific technological processes employed in their generation. Consequently, AI-generated evidence Sakshya Adhiniyam authentication processes are designed to integrate these new forms of data into existing legal structures.

The mechanism for handling electronic evidence, including AI-generated content, is primarily governed by the certificate regime outlined in Section 63 of the BSA, which succeeds Section 65-B of the Evidence Act, 1872. Landmark Supreme Court rulings have shaped the approach to electronic evidence. In *Anvar P.V. v. P.K. Basheer*, the Court determined that secondary electronic evidence is inadmissible without a certificate detailing the device and method of production. Crucially, this ruling did not deem electronic evidence inherently suspect but rather highlighted a procedural deficiency in uncertified submissions.

A subsequent decision by a three-Judge Bench in *Arjun Panditrao Khotkar v. Kailash Kushanrao Gorantyal* reaffirmed the mandatory nature of this certificate while introducing flexibility. This ruling permitted the certificate's production at a later stage and empowered courts to issue one if a party was unable to do so. These judicial pronouncements, taken together, do not create a barrier to new forms of digital evidence, including those from generative AI documentary evidence India, but instead establish a clear procedural pathway for their inclusion. The focus for India electronic evidence authentication has thus shifted from a general question of admissibility to the specific requirements of proving the integrity and origin of the data.

While the legislative structure places no categorical restriction on the acceptance of AI-generated evidence, the true evidentiary challenge lies beyond mere admissibility. The critical issue is authentication, which poses a distinct and more complex problem. Admissibility addresses a threshold question: whether a particular category of material is legally permitted to be introduced into the record. Authentication, conversely, asks a continuous question: can this specific item be trusted to accurately represent what it claims to be?

Indian evidentiary reforms since the *Anvar* case have shown increasing liberality regarding admissibility, yet the question of authentication has largely remained unaddressed in a specific context for AI. This asymmetry is central when AI-generated material is presented to a tribunal, as the two questions do not scale uniformly. Unlike a photocopy, which has a physical counterpart, a generative AI system can produce information that appears identical to authentic documents without any direct physical origin. This distinction highlights the need for a nuanced approach to AI evidence provenance reliability India, focusing on three critical aspects: provenance (the origin and generation process), reliability (the trustworthiness of the AI system), and independent corroboration of the AI's output. For AI-generated evidence, the primary legal hurdle has decisively shifted from mere admissibility to the rigorous demands of authentication.

Lawyers and compliance officers must understand that while the Bharatiya Sakshya Adhiniyam does not bar AI-generated evidence, the focus has shifted from mere admissibility to rigorous authentication. Practitioners need to develop strategies for establishing the provenance, reliability, and independent corroboration of AI-generated documents to ensure their acceptance and evidentiary weight in Indian courts, particularly in light of the Section 63 BSA certificate regime.

Source: Original reporting via legal analysis of the Sakshya Adhiniyam