Article 21 remains a divided promise in the Supreme Court’s bail jurisprudence

Should “liberty” under Article 21 of the Constitution be subject to the identity of the accused (young/old, politician/political science student, minority/non-minority) or should it remain an unqualified guarantee, as the framers of the Constitution intended when they drafted the sentence “No person shall be deprived of his life or personal liberty except according to procedure established by law”?

The Supreme Court’s recent bail jurisprudence under the Unlawful Activities (Prevention) Act, 1967 (UAPA), and the Prevention of Money Laundering Act, 2002 (PMLA) reveals deep inconsistencies in both outcomes and reasoning. The same statutory provisions, and often the same factual matrices, have produced contrary results from different benches, sometimes even the same bench.

The Court appears to take a subjective interpretation of Article 21, tilting on who the accused is, which bench hears the case and the vagaries of the political moment. Based on my analysis of the reasoning of the Court in 37 bail matters under the UAPA and PMLA, I argue that the jurisprudence exposes a pattern of contradictions, where liberty turns on rhetoric, subjectivity and selective readings of Article 21. I then propose a prototype framework for the grant of bail and apply it to Umar Khalid’s ongoing petition. 

A pattern of inconsistency 

In February 2024, a Bench of Justices M.M. Sundresh and Aravind Kumar, in Gurwinder Singh v State of Punjab, observed that a “mere delay in trial pertaining to grave offences cannot be used as a ground to grant bail.” The Bench went on to deny bail to an accused who was in custody for five years on a UAPA charge. Six months later, the same Bench took an opposite view in Mohd. Enamul Haque v Directorate of Enforcement: it held that a PMLA accused was entitled to bail due to his prolonged incarceration and a delay in trial. 

Both cases involved long custody and long delay in trials. But where the Court considered delay in trial as irrelevant in Gurwinder Singh, it treated it as decisive in Mohd. Haque. 

The pattern repeats across benches. In State (NCT of Delhi) v Raj Kumar @ Lovely (2024), a Bench of Justices Vikram Nath and Rajesh Bindal cancelled the Delhi High Court’s bail to a UAPA accused, warning that terrorism should not be “taken so lightly.” Yet, just three months later, in Shoma Kanti Sen v State of Maharashtra (2024), a Bench comprising Justices Aniruddha Bose and A.G. Masih held that constitutional courts are “not strictly bound by the prohibitory provisions” of UAPA and must apply Article 21 when incarceration becomes excessive. 

The contradictions become starker when considering Union of India v Barakathullah (2024). Here, a Bench of Justices Bela Trivedi and Pankaj Mithal cancelled the bail granted to members of the Popular Front of India, holding that ‘national security’ imperatives and the gravity of allegations justified continued detention. In this case, the accused had been in custody for a year-and-a-half. 

There is a similar divide under the PMLA. In cases concerning Aam Aadmi Party leader Arvind Kejriwal and then Bharat Rashtra Samithi leader K. Kavitha, the Court held that stringent bail conditions under Section 45 of the PMLA do not extinguish liberty under Article 21. Yet in Union of India v Kanhaiya Prasad (2025), a Bench of Justices Trivedi and P.B. Varale castigated a High Court for granting bail in a money laundering case, insisting that the approach of the courts cannot be “casual or cursory”. 

The same statute, the same Constitution, but two parallel universes of liberty. The law, in effect, changes with the bench. This contradiction is not merely doctrinal. It suggests that the Court’s approach to bail is shaped less by precedent than by perception. The “rule of law” has given way to what one might call the “rule of Bench”. 

Perhaps more troubling than inconsistency is selectivity. Article 21, the “soul of the Constitution”, is invoked vigorously for some and sparingly for others. The Court has expansively interpreted Article 21 to override statutory restrictions when the accused is a journalist (Prabir Purkayastha v State (NCT of Delhi)), a political leader (Arvind Kejriwal, K. Kavitha, Manish Sisodia v ED), or a senior citizen (Humayun Suleman Merchant v ED). It has spoken eloquently of liberty as “sacrosanct” and of the need to prevent incarceration from becoming punishment. Yet, when the accused is a young man charged under UAPA, the “sacrosanct” application is eclipsed by the rhetoric of “national security”, as seen in Raj Kumar @ Lovely or Barakathullah. 

There can be a contradictory reading within the same bench. A Bench of Justices Pankaj Mithal and Ahsanuddin Amanullah delivered a split verdict in politician Tahir Hussain’s bail petition. Hussain had been in jail for five years at the time. Justice Mithal dismissed his petition, stating that interim bail was not permissible for election campaigning. Justice Amanullah, however, emphasised Article 21 and long custody while noting that the gravity of allegations alone cannot justify the denial of bail. 

Interestingly, a few months ago, a bench comprising Justices Sanjiv Khanna and Dipankar Datta granted interim bail to then Chief Minister Arvind Kejriwal in a PMLA case to enable him to campaign for the General Election. The contrast between the orders in the Arvind Kejriwal and Tahir Hussain cases, one denying and the other affirming interim liberty for election campaigning, highlights the Court’s internal incoherence in applying Article 21.

This selective constitutionalism (or ‘curious constitutionalism’, in the coinage of criminal lawyer Abhinav Sekhri) corrodes both “equality” under Article 14 and “liberty” under Article 21. It creates a hierarchy of rights, robust for some, fragile for others. The Court’s own language oscillates: in Sheikh Javed Iqbal v State of Uttar Pradesh (2024), it declared that “a constitutional court cannot be restrained by restrictive statutory provisions if Article 21 is violated” but in Barakathullah, it pronounced that “national security is always of paramount importance,” effectively subordinating liberty to the executive’s claim of threat.

Such selectivity percolates down the judicial hierarchy. The consequence is a fragmented jurisprudence in which High Courts cite whichever precedent aligns with the desired outcome. For instance, in at least two mechanical orders from September 2025 alone, the Punjab and Haryana High Court has referenced Sheikh Javed Iqbal while granting bail to UAPA-accused (Sukhjinder Singh @ Bittu v State of Punjab and Ashish Kumar v State of Punjab). 

In contrast, the Madhya Pradesh High Court, in Washid Khan v State of Madhya Pradesh (2025), applied Barakathulla to reject bail. In Joginder Singh v NIA, the Delhi High Court, invoked Gurwinder Singh to deny bail. This selective borrowing of precedent, shaped by the Supreme Court’s own inconsistencies, has produced a jurisprudence of uncertainty, wherein the promise of equal protection and liberty becomes contingent on the story and citation used, not on the standard set by the Constitution. 

Formulating the ‘Khalid Test of Liberty’

The pending bail plea of Umar Khalid, a doctoral scholar and activist, epitomises the fault lines of India’s bail jurisprudence. Khalid was charged under the UAPA for alleged involvement in the Delhi Riots conspiracy. He has sought bail from the Delhi High Court and the Supreme Court on multiple occasions since September 2020. In September 2025, the Delhi High Court rejected his plea stating that there were “grave” allegations against Khalid and that bail cannot rest on delay alone. Meanwhile, the trial has barely progressed. Hundreds of witnesses remain unexamined.

Khalid has appealed the High Court’s decision in the Supreme Court. His case could be the crucible in which the Supreme Court’s rhetoric on liberty will be tested again. Multiple cases cited in this article have recognised that constitutional courts can override stringent provisions in favour of liberty. The question is whether those principles apply equally to Khalid. 

The ‘Khalid Test of Liberty’ can be framed as a constitutional formula emerging from the Supreme Court’s own precedents:

If applied consistently, the Khalid Test could transform bail from a discretionary indulgence to a constitutional entitlement. It would align the Court’s jurisprudence with its moral responsibility to ensure that punishment does not precede conviction. Khalid’s case, therefore, is not merely about one man’s freedom; it is a mirror held up to the Court.  

Reclaiming the promise of Article 21 

Today, the Supreme Court stands at a crossroads in its engagement with liberty. Its recent jurisprudence under UAPA and PMLA reflects both courage and confusion.The larger tragedy is not that bail is denied, but that the reasons for denial are inconsistent and identity-driven. 

When similar facts produce opposite results, the constitutional promise of Article 21 is reduced to a matter of luck. To reclaim that promise, the Court must anchor its bail jurisprudence in principled consistency. A restructured approach, grounded in a constitutional presumption of liberty, uniform criteria for delay and transparency in reasoning can restore coherence to the law. Above all, the Court must reassert that threats to “national security” and “financial systems” are not blank cheques to detain indefinitely. 

The true test of constitutionalism lies not in how the Court protects the powerful, but in how it guards the powerless. Until the Court embraces that equality of freedom, Article 21 will remain a divided promise,  majestic in text, but uncertain in life.

Sarthak Gupta is a New Delhi-based advocate and Researcher at Columbia University, New York, working on a project on free speech, media laws and censorship. He has previously served as Judicial Law Clerk to Justices Sandeep Mehta and Rajesh Bindal at the Supreme Court. He was not involved in the cited cases in any capacity, and all views expressed are his own, not those of the Court.