New criminal codes fall short on reforms on several counts

The monsoon session of parliament saw some eyebrow-raising legislative activity on its final day. Three new bills, which curiously featured only in a supplementary business list for the final day, were introduced in the Lok Sabha to overhaul criminal laws in India, purportedly to undo the “colonial” outlook of the criminal justice system. 

An old wine in a new bottle may be an apt metaphor for these bills. The bulk of their provisions seem paraphrased or to replicate, rather than shun, colonial servility. Not only that, but a crucial legislation seems to have escaped this supposed purge of colonialism -- the Police Act, 1861, arguably antiquated and in need of serious upgrades. 

Thus, the mere symbolic renaming of laws has failed to make India’s outdated criminal laws more contemporaneous. Specifically, given the massive technological innovation and deployment that presently facilitates different stages of law enforcement, it would have been meaningful for these new bills to deal with these ubiquitous technologies within their provisions. In particular, the use of biometric systems like facial recognition (FR), which has been vociferously assailed by privacy and civil rights advocates both in India and abroad, should have been addressed. 

Facial recognition is being increasingly used in policing across 20 states in India. Its usage has led to arrests, detentions, and random scanning of citizens nationwide. In fact, 137 people were arrested after the Delhi riots in 2020 based on FR matches. While the intended and actual use of FR is still very much shrouded in mystery, so is its legal basis. For instance, there is no law that permits or regulates the use of FR in everyday policing. There is also nothing in the existing CrPC to indicate that a FR match can be used as a basis for an arrest, search or even a preventive arrest. With over 8 million individuals preventively arrested every year, and increasing use of FR by the police, we are perilously close to dismantling the procedural due-process requirements guaranteed in Article 21 of the Constitution. 

The proposed Bhartiya Nagrik Suraksha Sanhita (BNSS) would have done well to address these concerns and lay down the due process requirements for use of FR. The BNSS must also have addressed the question on whether an FR match would be enough to inspire suspicion or provide reasonable grounds for arrest, search, or seizure. In fact, a discussion paper published by the NITI Aayog last year recommended legislative and policy reforms to ensure the deployment of FR is in accordance with responsible Artificial Intelligence (AI) principles. Akin to the proposed European Union AI Act, the BNSS could even prescribe a judicial oversight (say, by requiring a judicial warrant) as a prerequisite, to justify the use of FR systems in exigent and exceptional circumstances. 

Is FR admissible evidence?

Beyond legitimising and regulating the use of FR, there is another critical concern which remains unanswered. At present, there is a lack of recognition and admissibility of algorithmic output as evidence in a criminal (or civil) proceeding in Indian courts. The law of evidence in India has a limited recognition of electronic evidence, which only covers a digital version of physical evidence (say, an email instead of a physical letter or document). Algorithmic output is much more complex. It’s non-transparent, often inscrutable, and involves complex data-processing and syntheses which ultimately deliver a particular output (say, a match score or percentage in the case of FR). The existing definition of “electronic evidence” cannot by any stretch be construed to include such an output.

Further, there is also a strong argument against reliance on algorithmic output (of, say, an FR system) in identifying an accused and charging them in prosecution. This has already happened in Delhi, where in the Jahangirpuri riots case, the prosecution reportedly relied on FR matches and filed them in its chargesheet. Such usage is highly untenable under the existing legal framework. Not only is there a legal grey area on how to decipher FR algorithmic output to identify an accused, but given the limited understanding of electronic evidence, use of FR outputs in criminal prosecution is suspect and assailable. 

The Bharatiya Sakshya Bill has patently missed this opportunity to lay down rules on ‘relevancy’ of algorithmic outputs in criminal prosecutions and standards for their admissibility and reliability. It retains the current understanding of electronic evidence, thus rendering the use of FR or other predictive algorithms in policing vulnerable to legal challenges. 

Beyond procedural deficiencies, given that facial recognition can have a chilling effect on free speech and informational privacy, it would be necessary for the new criminal codes to adhere to the principle of proportionality. In addition to judicial warrants, as recommended above, both the BNSS, as well as the Bharatiya Nyaya Sanhita (BNS) must prescribe accountability measures. The BNSS should lay down procedures for deployment of FR. Chapter V and XII of the BNSS must also categorically provide for situations and the manner in which FR may be used to effect an arrest, search or seizure. Any arrest, detention, search, or seizure made based on an illegal or improper deployment of FR technology must be made punishable under the BNS. Provisions related to false evidence and false charges under Chapter XIV of the BNS may also be suitably amended to recognise malicious prosecutions with an intent to target and police specific communities.

As we advance in a more technologically equipped society, it is imperative for the law to enact ex-ante measures of accountability and guardrails, rather than play catch up. That is all the more necessary for criminal justice reforms! 

(Ahmad is Senior Resident Fellow, Criminal Justice team, and Jauhar leads the Centre for Applied Law and Technology Research, at the Vidhi Centre for Legal Policy) 

(This is the eleventh article in a DH-Vidhi Centre for Legal Policy series on the proposed new criminal law codes)