AI-Driven Investigative Technologies And Their Evidentiary Validity
Mohit Kumar Gupta, Advocate, Delhi High Court (Enrolment No – D/10507/2023) LLM, Jamia Millia Islamia, New Delhi (2023-2025)
ABSTRACT
The integration of Artificial Intelligence into India’s criminal justice administration has accelerated substantially, with law enforcement agencies deploying Automated Facial Recognition Systems (AFRS), probabilistic genotyping, predictive policing algorithms, and Natural Language Processing tools across investigative and prosecutorial processes. While these technologies promise enhanced investigative efficiency, evidentiary accuracy, and resource optimisation, their deployment has outpaced the development of a coherent legal and regulatory framework capable of governing their use within a rights-respecting constitutional order. This paper critically examines the evidentiary validity, constitutional implications, and regulatory lacunae surrounding AI-driven investigative technologies in the Indian context. It evaluates the adequacy of the current legal situation in addressing the admissibility, authenticity, and reliability of AI-generated evidence. It identifies the doctrinal challenges posed by the “black-box” opacity inherent in machine-learning systems. Through comparative engagement with the EU AI Act (2024), the US Daubert standard, and Singapore’s statutory presumption model, the paper identifies international best practices and evaluates their transposability to the Indian legal context. The paper concludes by proposing a dedicated statutory framework for AI in criminal justice, encompassing mandatory algorithmic impact assessments, judicially recognised testing standards for AI-generated evidence modelled on the Daubert framework, an independent AI oversight authority, and constitutional compliance protocols ensuring that AI serves as an instrument of justice rather than an engine of institutional arbitrariness.
