When Seeing Is No Longer Believing: Rethinking the Admissibility of AI-Generated Evidence in Indian Courts

Author

Nidhi Mishra

Abstract

 The increasing use of artificial intelligence (AI) to generate, manipulate and reproduce audio, video, images and documents has created significant challenges for the law of evidence. In India, the growing availability of deepfakes and other AI-generated content raises important questions concerning the authenticity, reliability and admissibility of digital evidence before courts. The legal framework governing electronic and digital records must now respond to evidence that may appear genuine while having been artificially created or substantially altered. This paper examines the admissibility of AI-generated and AI-manipulated evidence in Indian courts, with particular focus on the Bharatiya Sakshya Adhiniyam, 2023 and the principles governing electronic evidence. It analyses the evidentiary concerns arising from authenticity, integrity, attribution, chain of custody and the possibility of manipulation through generative AI. The paper further examines the role of expert evidence and technological methods of verification in assisting courts in determining whether AI-generated material can be relied upon. It argues that the existing framework provides a foundation for dealing with digital evidence but may face practical difficulties where sophisticated AI manipulation makes conventional methods of authentication inadequate. The paper concludes by proposing a technology-neutral and reliability-based approach that balances the evidentiary value of emerging digital technologies with the requirements of a fair trial and procedural justice.

Keywords

AI-Generated Evidence, Admissibility of Electronic Records, Bharatiya Sakshya Adhiniyam (BSA), 2023, Bharatiya Sakshya Adhiniyam (BSA), 2023, Hash Value Verification

Full Text:

Download Paper PDF

References

The increasing use of artificial intelligence in the creation and manipulation of digital content has fundamentally altered the evidentiary challenges faced by courts. Audio recordings, videos, images and other digital material can now be generated or modified with a degree of realism that makes traditional methods of visual or auditory assessment increasingly insufficient. The central difficulty is therefore no longer merely whether a digital record exists, but whether the record accurately represents the event, communication or person that it purports to represent.

The analysis undertaken in this paper demonstrates that the Bharatiya Sakshya Adhiniyam, 2023 (‘BSA’) provides a significant statutory foundation for dealing with emerging forms of digital evidence. Sections 61–63 recognise electronic and digital records and establish the statutory framework for their proof. The technology-neutral character of these provisions is advantageous because it allows the law to accommodate technological developments without requiring a separate evidentiary regime for every new form of digital technology.[38] However, the absence of provisions specifically addressing AI-generated and AI-manipulated evidence creates important practical and doctrinal uncertainties.

A central finding of this study is that admissibility and authenticity must be treated as distinct inquiries. Compliance with the statutory requirements governing electronic evidence does not necessarily establish that the content of an electronic record is genuine. An AI-generated video may satisfy the formal requirements for production as an electronic record while nevertheless falsely depicting an event. Similarly, an AI-generated voice recording may accurately reproduce an individual's voice without establishing that the individual actually made the alleged statement. The evidentiary inquiry must therefore extend beyond the formal admission of the digital record to its provenance, integrity, authenticity and attribution.

The jurisprudence developed under the earlier Indian Evidence Act, particularly Anvar PV v PK Basheer and Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal, demonstrates the Supreme Court's emphasis on statutory compliance and reliable proof of electronic evidence.[39] Although these decisions arose under section 65B of the Indian Evidence Act, their broader emphasis on authentication and reliability remains relevant when courts confront technologically sophisticated forms of digital evidence under the BSA.

The study further establishes that expert evidence and digital forensic examination will assume increasing importance in cases involving suspected AI manipulation. Section 39 of the BSA provides a statutory basis for expert opinion concerning science, technology and information stored or transmitted through electronic or digital systems.[40] Nevertheless, expert evidence should not be treated as conclusive merely because it is based upon technological analysis. Courts must consider the methodology employed, the reliability of the forensic process and the limitations of AI-detection technologies. The determination of authenticity should consequently be based upon the totality of the available evidence rather than upon a single technological indicator.

Another important finding is the significance of provenance and chain of custody. Where digital evidence is capable of being repeatedly copied, modified and redistributed, establishing its source and history becomes essential. The preservation of original files, documentation of the manner in which evidence was obtained and stored, and appropriate forensic examination can assist courts in distinguishing genuine evidence from manipulated or fabricated material.

The paper therefore concludes that the BSA is not fundamentally inadequate for dealing with AI-generated evidence, but its existing provisions are not, by themselves, sufficient to resolve every authenticity dispute created by sophisticated AI manipulation. The legislation provides the basic evidentiary framework, while courts will need to develop appropriate methods of applying existing principles to new technological circumstances. A separate rule automatically excluding AI-generated evidence would be undesirable because the use of artificial intelligence does not necessarily make evidence false or unreliable. Conversely, treating digitally generated material as reliable merely because it satisfies the formal requirements for electronic evidence would create a serious risk of wrongful findings based upon fabricated material.

The appropriate approach is therefore a technology-neutral, reliability-based framework. Courts should examine the source, provenance, integrity, authenticity and attribution of disputed digital material and, where necessary, require expert or forensic verification. The mere possibility of AI manipulation should not be sufficient to exclude otherwise relevant evidence, just as the mere existence of a technically valid electronic record should not prevent the court from investigating credible allegations of manipulation.

Ultimately, the challenge presented by AI-generated evidence is not whether artificial intelligence should be accepted or rejected by the law of evidence. The real question is whether the legal system can develop reliable methods for distinguishing authentic digital evidence from convincingly fabricated digital evidence. The BSA provides a foundation upon which such an approach can be developed, but consistent judicial standards, improved forensic capabilities and appropriate procedural safeguards will be necessary to preserve the integrity of digital evidence.

The future of Indian evidence law should therefore not be based upon technological suspicion, but upon verifiable reliability. A balanced approach that combines statutory compliance, forensic examination, expert assistance, provenance verification and judicial scrutiny can enable courts to benefit from legitimate technological evidence while protecting accused persons and other litigants from the dangers posed by sophisticated synthetic media. Such an approach would best reconcile technological innovation with the fundamental objectives of fair trial, accurate fact-finding and procedural justice.

Share your valuable work from Social Media Buttons