Krishna Dube

Abstract: The Bharatiya Sakshya Adhiniyam, 2023 retains the near-absolute exclusion of confessions made to, or in the custody of, the police, reflecting concernings regarding custodial coercion. This article argues that the categorical exclusion is structurally insufficient because it treats all custodial confessions alike, without permitting judicial scrutiny of statements recorded under verifiable safeguards. It proposes a structured framework of safeguarded admissibility based on audio-visual recording, legal consultation, documented custodial safeguards, and judicial assessment of voluntariness.

INTRODUCTION

Confessions occupy a central place in criminal proceedings because they directly implicate the accused and often shape both investigation and trial. Indian evidence law has historically treated such statements with extreme caution, especially when they are recorded in police custody. The Bharatiya Sakshya Adhiniyam, 2023 (‘BSA’) retains this approach by placing a near-absolute bar on confessions made to, or in the custody of, the police, subject to the limited discovery exception. This reflects a long-standing concern that statements obtained in custody could be influenced by pressure, inducement, or fear of authority, and therefore cannot be relied upon without strict safeguards.

The risks associated with custodial statements are not merely limited to instances of coercion. Various factors, such as the manner in which the questioning takes place, the conditions of detention, and the control exercised by the police over the suspect during custody, all shape how statements are made and recorded. A rule that treats all confessions in the same way, without having regard to these differences, assumes that the same risks arise in each and every case. This leaves no room to distinguish between statements obtained under undue pressure and those recorded under conditions permitting fair scrutiny.

The article argues that the BSA’s categorical exclusion is structurally insufficient because it fails to engage with the process of interrogation rather than merely its outcomes. Also, the proposed reforms draw on comparative experience not to simply mirror foreign models in India, but to identify procedural instruments that are compatible with India’s existing infrastructure. The article asks whether the BSA addresses these concerns and argues that a structured admissibility framework based on clear safeguards and judicial scrutiny would regulate police confessions in a much more coherent manner than categorical exclusion.

THE EXCLUSION OF POLICE CONFESSIONS IN INDIA:LEGAL STRUCTURE, SCOPE, AND LIMITATIONS

A. Statutory Framework

The BSA sets out a consistent regime with regard to how confessional statements are to be treated in criminal trials. Section 23(1) provides that no confession made to a police officer shall be proved as against a person accused of an offence, while Section 23(2) extends this exclusion to confessions made in police custody unless recorded in the immediate presence of a Magistrate. When read together, these provisions establish a general rule of inadmissibility for custodial confessions, subject to the limited exception of judicially supervised recording. However, this creates a strange gap. The police can use the information to find evidence, and that evidence would be admissible in court, but the statement which led to the finding cannot be used at all.

The statutory framework separates the investigative process from the recording of evidence, assigning the latter to the Magistrate as a safeguard against undue influence. The magistrate mechanism under Section 183 of the Bharatiya Nagarik Suraksha Sanhita, 2023 (“BNSS”) is a post hoc process. Custodial questioning takes place when the investigation is ongoing, during which evidence is undiscovered, accomplices are unidentified, and each lead is time-sensitive. A statement made at that point directs the course of the investigation. Further, magistrate confessions are regularly retracted at trial and require corroboration. The discovery exception already admits the fruit of custodial statements. If the information is usable, the statement produced should be made admissible under strict safeguards.

Indian courts have consistently recognised the potential of there being coercion in a custodial interrogation. The Supreme Court in Nandini Satpathy v P.L. Dani held that Article 20(3) of the Constitution applies also during an interrogation, not only at trial. Further, the Court also suggested that lawyers may be present during the questioning, something which remains unlegislated but aligns with the proposed framework’s requirement of legal consultation. In D.K. Basu v State of West Bengal, the Court issued mandatory guidelines for arrest and detention, which included requirements for custody registers as well as medical examinations. In the case of Selvi v State of Karnataka, the court identified voluntariness as being the heart of Article 20(3). Additionally, the Court reaffirmed the exclusionary rule in Tofan Singh v State of Tamil Nadu, deciding that officers under special statutes are to be treated as police officers for confession purposes.

The 69^th Law Commission Report of 1977 recommended making confessions to senior police officers admissible under certain conditions. The 185^th Law Commission of 2003 proposed expanding Section 24 to explicitly include coercion torture, and violence as grounds for exclusion. This shows that even India’s own law reform bodies have endorsed conditional admissibility of custodial confessions.

B. Custodial Reality and the Limits of Formal Exclusion

The statutory structure of the BSA rests on the premise that custody creates a uniform risk of undue influence. In practice, however, custodial questioning does not always operate in the same manner, but through different methods that affect the reliability of the resulting statement. Interrogation models that proceed on an assumption of guilt and rely on sustained psychological pressure have been proven to produce a higher incidence of false confessions. On the other hand, structured information-gathering methods are related to a higher rate of true confessions and lower false admissions. The reliability of a confession is thus shaped significantly by how the interrogation is conducted, and not just by the fact that it was made in police custody.

Influence in custody also operates through factors that may not be readily visible in the formal record. The decision-making of an accused is often shaped by the interaction between the pressure of the situation and individual vulnerability, including cognitive limitations, suggestibility, and psychological susceptibility. These factors may affect responses even where no threat or inducement may visibly be present. Judicial standards that focus primarily on demonstrable external pressure risk overlooking these subtler influences, with the result that forms of constraint that do not take the shape of overt coercion may escape scrutiny altogether. On the face of it, this might seem to be supporting exclusion. However, there is a difference between vulnerability arising from the custodial environment, which can be controlled by proper safeguards, and pre-existing vulnerability, which needs to be assessed on a case-to-case basis. A categorical exclusion addresses neither; it excludes the accused’s voluntary statement while offering no protection.

Data shows that India has averaged around 160 custodial deaths per year in the last 5 years. It suggests that the exclusion rule does not really deter custodial violence; it merely makes the confession inadmissible without addressing the underlying conditions or environment. Recently, in the Nithari case, the accused’s confession was recorded after nearly sixty days of police custody without access to any legal assistance. The Supreme Court refused to treat it as reliable. This illustrates that judicial scrutiny of custodial conditions is already occurring.

The conditions of custody further affect how individuals respond to questioning. Statements may be made solely to bring an end to the questioning or to protect someone else. Choices to deny allegations or to remain silent may vary with perceptions of evidentiary strength and access to legal advice. The proposed reform addresses this by making legal consultation a mandatory precondition for recorded interrogation, as in Miranda v. Arizona. In this sense, custodial statements may reflect responses to the pressures and structure of interrogation rather than a reliable and accurate account of facts. Once obtained, confession evidence assumes considerable significance within the criminal process, shaping the direction of the investigation as well as the conduct of the prosecution.

It has been seen from jurisdictions that recognise exclusion of illegally obtained confessions that formal prohibition alone does not ensure effective regulation. Questions arise in determining what constitutes illegality, how the burden of proof is to be discharged, and how courts are to assess the circumstances in which statements are obtained.  Effective oversight has instead depended on procedural mechanisms permitting judicial inquiry into the conditions of the interrogation and the manner in which a statement was obtained.

Against this backdrop, BSA applies a general bar to custodial confessions without differentiating between interrogation practices that carry a heightened risk of false admission and those conducted under documented safeguards. It also does not provide a structured basis for judicial examination of custodial conditions. The issue, therefore, lies not in recognising the risks inherent in custody, but in whether a uniform rule of exclusion is capable of addressing the varied ways in which statements are produced. Comparative approaches illustrate an alternative model in which admissibility is linked to identifiable safeguards and is subject to judicial scrutiny.

ADMISSIBILITY OF POLICE CONFESSIONS IN OTHER JURISDICTIONS: A COMPARATIVE ANALYSIS

This section examines how selected jurisdictions regulate the admissibility of police confessions through identifiable safeguards and the judicial scrutiny of interrogation practices. It analyses the statutory and judicial frameworks in England and Wales, the United States (‘U.S.’), Canada, and China, and illustrates how they address the risks related to custodial interrogation without resorting to categorical exclusion.

A. England and Wales

In England and Wales, the admissibility of police confessions is governed primarily by the Police and Criminal Evidence Act, 1984 (‘PACE’). Section 76 of the Act mandates the exclusion of a confession where it is obtained by oppression or in circumstances likely to make it unreliable, and places the burden on the prosecution to prove compliance beyond a reasonable doubt. “Oppression” has been interpreted to require improper or unjust conduct by investigators rather than the suspect’s internal distress alone. In R v Fulling, the Court of Appeal held that emotional pressure felt by the accused did not amount to oppression in the absence of wrongful police conduct.

Section 78 of the Act permits the court to exclude evidence where its admission would adversely impact the fairness of the trial. In R v Mason, the court exercised this discretion after investigators falsely claimed to have incriminating evidence during the interrogation. This misrepresentation was made not only to the accused but also to his solicitor, and the solicitor’s advice to cooperate was in turn based on that falsehood. The Court described this as “hoodwinking both solicitor and client,” rendering the confession unfair. Thus, English law permits police confessions in principle but subjects them to close judicial scrutiny of reliability and fairness, assessed in light of the manner in which the statement was obtained.

b. U.S.

In the U.S., the admissibility of confessions is governed by constitutional guarantees under the Fifth and Fourteenth Amendments. In Brown v Mississippi, the Supreme Court held that confessions obtained through physical violence violate due process and are inadmissible. The modern framework was shaped by the landmark case of Miranda v Arizona, which requires that suspects must be informed of their right to remain silent and to consult counsel before custodial interrogation. Statements obtained without such safeguards are generally excluded as evidence.

Subsequent decisions have clarified the limits of these protections. For instance,  in Missouri v Seibert, the Court addressed the practice of deliberately interrogating a suspect without warnings, securing a confession, and then obtaining a repeated confession after administering Miranda rights. A plurality evaluated whether the later warning was effective in the context of the entire questioning, while a concurring opinion confined exclusion to situations where the initial unwarned interrogation was deliberately used to circumvent procedural safeguards. The decision demonstrates that admissibility not only depends on formal compliance but also on whether safeguards operate effectively in practice. Further, in Colorado v Connelly, it was held that a confession cannot be said to be involuntary in the constitutional sense unless coercive police conduct is present, even if the accused suffers from a mental illness. The American position thus permits police confessions to be admitted, subject to procedural safeguards and the absence of coercive state action.

c. Canada

Canadian law adopts a structured and context-sensitive approach to voluntariness. In the case of R v Oickle, the Supreme Court of Canada held that admissibility depends on an assessment of the totality of circumstances, including threats or promises, oppressive conditions of detention, the state of mind of the accused, and the use of police trickery. No single factor is determinative. The main point to be seen is whether the statement represents a free and informed choice on the part of the accused. Such a system allows courts to evaluate how a confession was obtained rather than excluding it only on the basis of custody. It demonstrates that police confessions can be regulated through a principled evidentiary inquiry without undermining protection against coercion.

d. China

Chinese law permits the use of confessional evidence, subject to statutory restrictions. Article 54 of the Criminal Procedure Law requires courts to exclude confessions obtained through torture or other illegal means. The Rules for Handling the Exclusion of Illegal Evidence in Criminal Cases supplement this framework by prescribing procedures for identifying and excluding such statements. They recognise that the effects of coercion may easily extend beyond the initial interrogation, and that subsequent confessions may also be excluded unless shown to be independent of earlier illegality. The Chinese model, therefore, combines formal exclusion with procedural mechanisms aimed at examining how a statement was obtained.

Across these jurisdictions, police confessions are not admitted without restriction. Their admissibility depends on compliance with procedural safeguards and judicial scrutiny of the circumstances in which the statement was obtained. The comparative analysis shows that the risks associated with custodial interrogation are addressed through regulated admissibility rather than absolute exclusion.

Comparative Lessons for the Indian Context

From England and Wales, the structured exclusion framework with the prosecution bearing the burden of proving voluntariness is directly applicable in India. Courts already conduct voir dire inquiries in certain contexts, and extending the same to confession voluntariness only requires a statutory mandate.

From the U.S., the Miranda framework is relevant, but formal compliance can mask circumvention. While adapting it in India, evidentiary consequences to non-compliance need to be attached, and not just procedural sanctions. The existing CCTV mandate and BNSS Section 176(3) provide for recording to verify that all safeguards were met before and during questioning.

The Oickle totality of circumstances approach from Canada is appropriate for India. There already exists multi factor reasoning in other evidentiary contexts and introducing such a system does not require setting up of new institutions. It only requires judicial training and a codified list of factors.

From the Chinese system, India can adopt the practice of courts examining the sequence of interrogation in order to establish whether later statements remain influenced by earlier coercion, rather than treating them as independent merely because they have been recorded at a different point of time.

REFORMING THE BSA: A STRUCTURED FRAMEWORK FOR ADMISSIBILITY

The present rule excludes confessions made to the police in their entirety. While this position addresses the risk of coercion during interrogations, it does not take into account the circumstances in which statements are actually obtained. A more coherent approach would be to retain the safeguards against coercion while allowing courts to examine whether a confession was recorded under verifiable safeguards.

Firstly, custodial interrogation should be audio-visually recorded in a continuous manner from the beginning of questioning till it concludes. Although such a recording is provided in Section 176(1) of the BNSS, it does not make it mandatory or attach any evidentiary consequence to non-compliance. This proposal is not one without precedent. The Supreme Court in Paramvir Singh Saini v Baljit Singh mandated CCTV in all police stations with night vision along with audio-video capability. In 2025, the Court initiated suo motu proceedings upon finding incomplete compliance, with non-functional cameras cited in multiple cases of custodial death. This history of non compliance is not an argument against reform, but is one which advocates for attaching evidentiary consequences. The present CCTV mandate is a transparency measure without evidentiary weight.

Under the proposed reform, recording must cover the entire interaction, including all intervals, and form part of the investigation record. Where the prosecution relies on a confession, the recording must be produced before the court. Where interrogation has not been recorded without a verifiable reason, the confession should not be relied upon as evidence. Continuous recording would provide an objective account of interrogation practices and allow courts to decide whether the statement was obtained through pressure, inducement, or improper questioning.

Secondly, the admissibility rule under Section 23 of the BSA should be conditioned through a narrowly framed proviso. It may provide that a statement made during an interrogation shall be admissible only if the interrogation was audio-visually recorded in its entirety, the accused had been informed of the right to consult a lawyer before questioning, and the statement was subsequently confirmed before a Magistrate within a specified period after interrogation. This would retain the existing concern regarding custodial influence while allowing courts to examine statements recorded under safeguards.

Thirdly, courts must conduct a structured inquiry to assess the voluntariness of the confession whenever the prosecution seeks to rely on a confession obtained during a police interrogation. They must take into account the duration and method of questioning, the conditions of detention, whether it was recorded, whether legal consultation was made available to the accused, and whether the accused showed signs of distress or vulnerability. The burden must be on the prosecution to prove that the confession was voluntary.

Assessment of distress could proceed through four mechanisms. First, where recording exists, the court may review footage to observe the accused’s behaviour, including distress, fatigue, confusion, or resistance. Second, mandatory medical examination of the accused before and after interrogation would help take note of the physical condition and injuries. Disparities between the findings may prove coercion. Third, where the accused raises a claim of cognitive or mental vulnerability, the court may order a forensic psychiatric assessment. Finally, the custody register should document the exact time at which the interrogation began and ended, the officers present during questioning, and the time at which the accused was presented before a Magistrate. Extended interrogation without recorded breaks further shifts the burden onto the prosecution.

Fourthly, safeguards must address the possibility of coercion taking place outside the recorded interrogation. Custodial questioning must take place only in designated interrogation rooms, and interrogators should not question the accused elsewhere except where unavoidable and documented in detail. The custody register should record the movement of the accused throughout detention, identity of officers having access, and all interactions during detention. A medical examination immediately prior to interrogation shall be recorded. Any statement made during the period between arrest and the commencement of recorded interrogation shall not be proved.

These safeguards give rise to a legitimate question: if a custodial confession requires such extensive scrutiny to be admissible in court, why not retain the current system of categorical exclusion and rely only on magistrate confessions? Under the existing rule, custodial interrogations occur as a matter of course. Suspects make statements, which, regardless of their voluntariness or reliability, cannot be proved at trial. This creates a very negative outcome. The police have no institutional reason to comply with procedural safeguards while interrogating the accused. Since their presence or absence does not affect admissibility, recording, legal access, and medical examinations are treated as being optional. The proposed framework incentives these safeguards. It provides that where the police follows the very safeguards that are mandated by existing law, the resulting statement may be admitted. Where they do not, the statement continues to remain excluded. This imposes no new substantive obligation, only attaches evidentiary value to the existing ones. Magistrate confessions, as earlier explained, remain available, however, they are post-hoc and are frequently retracted. A custodial confession, under verifiable safeguards, offers a procedurally fair alternative.

CONCLUSION

Confession evidence occupies a complicated position in criminal trials. While it may provide a direct account of the events, statements made in police custody remain particularly vulnerable to influence arising from the interrogation conditions and the authority which the investigators exercised over the accused. Indian evidence law has historically excluded confessions made to the police or in police custody. The BSA continues this position by retaining a near-absolute bar on police confessions, subject only to limited exceptions.

Although this rule reflects a legitimate concern about custodial coercion, it operates as a uniform exclusion that does not account for the circumstances in which confessions are obtained. Interrogations may vary in method, duration, and the safeguards present, and the reliability of a confession often depends on how the questioning is conducted and documented. Comparative experience from jurisdictions such as England and Wales, the US, Canada, and China shows that these risks are frequently addressed through procedural safeguards and judicial scrutiny rather than by excluding police confessions altogether.

The central issue, thus, lies not in recognising the risks associated with custodial confessions, but in determining how those risks should be addressed within evidence law. The reforms proposed in this article, including continuous audio-visual recording, documented custody conditions, access to legal consultation, and structured judicial inquiry into voluntariness, provide a basis for such regulation. These measures would allow courts to examine how a confession was obtained while retaining protection against coercion. In this way, the law can move from a rule of categorical exclusion to one that evaluates confession evidence through verifiable safeguards ensuring both reliability and fairness in trials.

Krishna Dube is a fourth-year B.A. LL.B. (Hons.) student at the National Law Institute University, Bhopal.