AI-Generated Evidence: Admissibility & Reliability in Indian Courts

The rapid entry of AI into daily operations from automated surveillance logs and predictive software to synthetic media and deepfakes is forcing Indian courts to rethink how truth is proven in a courtroom. While digital records are generally admitted under Section 63 of the Bharatiya Sakshya Adhiniyam, 2023, through procedural certification, AI outputs present a fundamental challenge: certifying the device that generated the data does not guarantee that the algorithm’s output is truthful, unbiased, or accurate. This article examines the critical divide between mere procedural admissibility and true substantive reliability. It highlights the dangers of algorithmic "black-box" opacity, hallucinations, and chain-of-custody gaps, arguing that Indian jurisprudence must establish stricter judicial testing standards, mandatory algorithm disclosures, and human-led scrutiny to protect the fundamental right to a fair trial under Article 21.