LAST WEEK, on May 19, 2026, a Division Bench of the Supreme Court, comprising Justices B.V. Nagrathana and Ujjal Bhuyan, in Syed Ifitkar Andrabi v. National Investigating Agency, Jammu, granted bail to an under-trial prisoner from Kashmir who had spent nearly six years in custody under the Unlawful Activities (Prevention) Act, 1967 (‘UAPA’), and Narcotic Drugs and Psychotropic Substances Act, 1985 (‘NDPS Act’), with over 350 witnesses still to be examined and no realistic prospect of his trial concluding within a visible horizon.
Andrabi is, in essence, a restatement of the liberty principle on bail under statutes like the UAPA, Prevention of Money Laundering Act, 2002 (‘PMLA’), and the NDPS Act. Yet, it also travels further to chastise the decisions of two coordinate benches (headed by Justice Aravind Kumar) that had not adhered to a binding precedent of a three-judge bench decision in Union of India v. K.A. Najeeb (2021). In its factual analysis, one can also witness a subtle emphasis that a Court, while adjudicating on bail, must conduct only a ‘prima facie’ assessment and not a ‘mini trial’.
Three days after Andrabi was delivered, on May 22, a Division Bench of Justices Kumar and P.B. Varale, while granting bail to Delhi riots conspiracy accused Tasleem Ahmed and Khalid Saifi referred to a larger bench the question of how Article 21 is to be applied against the statutory embargo on bail under Section 43(D)(5) of the UAPA. On May 20, the Delhi police had told the Court that following Andrabi, a larger bench reference may be needed given the existence of divergent views on grant of bail under UAPA.
In this essay, I focus on Andrabi’s reaffirmation of Najeeb and its criticism of recent decisions such as Gurwinder Singh v. State of Punjab (2024) and Gulfisha Fatima v. State (Govt. of NCT of Delhi) (2026) for diluting that precedent without formally overruling it. Secondly, I argue that Andrabi correctly addresses (but without theorising it) the growing phenomenon of what I term “admissibility-blind bail adjudication”, where courts accept chargesheet allegations and uncorroborated disclosure statements at face value while applying the prima facie standard under Section 43D(5) of the UAPA, even when such material is plainly inadmissible in evidence.
What happened in Andrabi
The present case concerns a government employee, Syed Ifitkar Andrabi, who was working at the Rural Development Department, Kupwara, Kashmir. After the abrogation of Article 370 of the Constitution in 2019, Syed was taken into preventive detention under the Jammu & Kashmir Public Safety Act, 1978, merely to ensure there was no breach of “law and order.” Two months before the High Court declared the preventive detention order illegal, the government itself revoked the order on June 25, 2020. Two months after revoking the order against him, an FIR was filed against Syed based on his confession that he had taken a narcotic substance, heroin, from another individual, which he himself could recover by merely pointing it out.
Later, he was charged under Sections 8, 21, 25, and 29 of the NDPS Act and Sections 17, 38, and 48 of the UAPA, read with Section 120B of the Indian Penal Code, 1861. The trial courts and then the Jammu & Kashmir High Court dismissed his bail petition noting that there was a prima facie case on narco-terrorism and “a strong suspicion” was established on “the strength of material on record of the Court.” (Para 22 of the High Court’s decision). It is pertinent to note that the High Court relied on Gurwinder to reject Syed’s ground of bail against long incarceration and delay in trial (Para 18).
The Supreme Court, last week, in an appeal against the High Court, granted bail to Syed on the ground that his continued incarceration for more than five years and nine months, coupled with the remote possibility of an early conclusion of trial, amounted to a violation of his fundamental right to personal liberty and speedy trial under Article 21 of the Constitution. The Court held that although Sections 37 of the NDPS Act and 43D(5) of the UAPA impose stringent restrictions on the grant of bail, such statutory embargoes cannot override constitutional protections where prolonged pre-trial detention becomes punitive in nature.
The Court also took note of the weak evidentiary foundation against the appellant, including doubts surrounding the admissibility of the alleged disclosure statement and recovery, absence of corroborative material linking him with terror activities, and the fact that similarly placed co-accused had already been enlarged on bail.
Najeeb and the Two ‘Gs’: Understanding the Judicial Art of Hollowing without Overruling
In Najeeb, a judgment authored by the current Chief Justice of India Surya Kant, a three-judge bench of the Supreme Court held that Section 43-D(5) of the UAPA does not per se oust the ability of Constitutional Courts to grant bail on grounds of violation of fundamental rights. The Court held that “rigours” of the statutory embargo would “melt down” where there was no likelihood of the trial concluding within a reasonable time and where the period of incarceration had already exceeded a substantial part of the prescribed sentence (Para 18). Thus, in Najeeb, the Supreme Court articulated a constitutional constraint on the operation of the statutory framework of Section 43-D itself.
Following Najeeb, a two-judge bench of the Supreme Court in Gurwinder, authored by Justice Aravind Kumar, disregarded the above constitutional constraint and formulated a twin-prong test to determine the question of bail in UAPA cases. In the said test, the Court first asks whether the accusation is prima facie true, and only if the answer favours the accused does the court move to conventional bail considerations. If the prima facie threshold is satisfied against the accused, bail becomes “absolutely impermissible” and is outside the scope of constitutionally guaranteed rights. As per Gurwinder, the legislative intent was that “bail must be rejected as a rule” (Para 20).
Two years later, in Gulfisha, authored by the same judge, the Court narrowed the constitutional constraint of Najeeb further. Although the Court acknowledged Najeeb, it misinterpreted the decision as merely a case-specific safeguard rather than a binding precedent (stare decisis). In doing so, Gulfisha diluted Najeeb, which had not treated prolonged incarceration as a discretionary or equitable consideration, but had expressly recognised that the rigours of Section 43D(5) must yield where continued detention violates Article 21. Gulfisha, however, characterised such a reading as producing an “interpretive absurdity” because it would allegedly neutralise special anti-terror legislation through the “mere passage of time” (Para 61). By recasting Najeeb as an exceptional or discretionary principle rather than a binding constitutional constraint, Gulfisha diluted the force of the earlier three-judge bench decision without formally overruling it.
On the other hand, Gulfisha referred to Gurwinder to state that the Court cautioned against the mechanical invocation of prolonged incarceration as a ground for bail, notwithstanding that it was categorically noted by the Bench in the Review Petition against Gurwinder that the decision ought to be construed on the facts dealt with by the bench.
In Andrabi, the Court directly (and in very thorough detail) characterised both of the above ‘Gs’ – Gurwinder and Gulfisha – as having made a clear departure from the ratio laid down in Najeeb without ever expressly disagreeing with it. The Court held that, as a principle of judicial discipline, a smaller Bench cannot dilute, circumvent, or disregard the ratio of a larger Bench. The Court also held that if a two-judge bench cannot agree with a three-judge ruling, the proper course is a reference to a larger bench.
Furthermore, the Court in Andrabi went on to criticise the formulation of the “twin-prong test” in Gurwinder in particularly incisive terms. The Court rightly held that the genesis of this test neither flows from the text of Section 43-D(5) nor from Najeeb, because Najeeb had expressly recognised that Section 43-D(5) stipulates “no more than another possible ground” for refusing bail, alongside the conventional tripod test governing bail adjudication. If the twin-prong test is accepted as the governing standard for deciding bail petitions, the Court noted, pre-trial incarceration would begin to acquire a “post-trial punitive character.” This is because no matter how long the accused languishes in custody, and no matter how remote the prospect of the trial concluding, bail would become permanently unavailable once the prima facie threshold is crossed. This is precisely the judicial constraint that Najeeb intended to avoid. Andrabi, therefore, reiterated the principle that “bail is the rule and jail is the exception,” even in cases arising under statutes such as the UAPA and the PMLA.
The most troubling consequence of Gurwinder’s misreading of Najeeb is visible in the manner in which it has shaped subsequent bail adjudication. By recasting Najeeb’s constitutional constraint as a narrow and exceptional principle, Gurwinder effectively created a doctrinal basis for coordinate benches of the Supreme Court (like the decision in Union of India v. Barakathullah (2024)), and High Courts to deny bail even in cases of prolonged incarceration and stagnant trials. For instance, before the Jammu & Kashmir High Court, Gurwinder has been cited in at least seven bail matters (according to the SCC Case Reference tool), including Andrabi, and each of those petitions were dismissed despite reliance being made on Najeeb.
What this pattern illustrates is that a later bench, without once formally disagreeing with the three-judge decision, can drain it of operative force until Article 21 protections survive only as citations in rejected petitions.
Is Admissibility-Blind Bail Adjudication a new dimension in UAPA Bail Crisis?
In Andrabi, the prosecution's case against Syed rested on the three elements: (a) a disclosure statement made by Syed before the Police, pursuant to which cash and heroin were allegedly recovered from accused No. 1's bedroom; (b) an “explanation memo” of phone contacts prepared based on Syed’s disclosure before the police showing alleged links to operatives of militant groups Lashkar-e-Taiba (‘LeT’) / Hizbul Mujahideen; (c) statements of protected witnesses and an approver implicating Syed in drug racketeering and terror funding; and (d) an “over-ground worker categorisation certificate” issued the day before the chargesheet was filed.
If we consider each element in turn, none of them is admissible in evidence at face. First, the disclosure statement is nothing but a confession made to a Police officer, which is inadmissible under Section 25 of the Indian Evidence Act, 1872 (now Section 23(1) of Bharatiya Sakshya Adhiniyam, 2023). There is no exception. It is not that the confession is weak evidence or requires corroboration; it simply cannot be received in evidence against the accused at trial. Second, the “explanation memo” of phone contacts is similarly tainted as it was prepared based on Syed explaining to the police the contacts appearing in his mobile phones, i.e., it is itself like a statement made by the accused before the police, thus, squarely prohibited under Section 25 of the Evidence Act.
What the Andrabi judgment does not address
Andrabi, for all its length (spread across a 102-page bail order), does not fully resolve every question it raises. The treatment of Gulfisha is notably restrained, as the bench says it has “serious reservations” on various aspects but declines to “join issue any further” with that bench, additionally noting that Gurwinder has already been explained in subsequent decisions. It casts doubts on the “twin-prong test”, but fails to set it aside or refer the test’s legality to a larger bench. However, a coordinate bench in Tasleem Ahmed did refer the question of how Article 21 is to be applied against the Section 43D(5) embargo to a larger bench, partially addressing this gap, leaving the question of the legality of the “twin-prong test” open, though one that the larger bench should now be compelled to address.
This reflects a careful way of declining to create a formal conflict while signalling judicial displeasure. Whether this is sufficient to prevent future two-judge benches from citing Gulfisha or Gurwinder remains to be seen.
A structural problem that Andrabi names but cannot fix is why trial courts and High Courts continued to deny bail on merits despite long incarceration, in cases where the primary prosecution evidence was arguably inadmissible from the first day. The answer likely is that courts below have learned, correctly, from the pattern of Supreme Court and High Court decisions, that the UAPA accusation is sufficient, and that interrogating the quality of the prosecution material at the bail stage invites the Watali rebuke about “mini-trials.”
The result is a system where the accused must wait for years until they can accumulate enough delay to trigger the Najeeb relief, rather than receiving bail on merits at the outset when the legal inadequacy of the case is apparent. In Gulfisha, the Bench, while denying bail to Umar Khalid and Sharjeel Imam, expressly noted that they both may approach after a year if the trial is not complete, indirectly stating that six years of incarceration is a prerequisite to grant bail.
Moreover, the tension between Watali, i.e., taking material at face value, does not determine admissibility at the bail stage, and the implicit recognition in Andrabi that plainly inadmissible material should at least be identified as such has also not been resolved doctrinally. A larger bench, now, will need to draw this line more explicitly, not as a licence for detailed evidentiary evaluation at the bail stage, but as a minimum threshold requirement that "prima facie material" relied upon to deny bail be at least facially receivable in evidence, that it belongs, as a matter of law, to the category of things that can be proved.
Whether the Court's clarification in Andrabi will hold, whether Najeeb will remain good law in the face of future two-judge attempts to distinguish it, and whether the State's immediate move for a larger bench reference will succeed in recreating the uncertainty that Gurwinder and Gulfisha had exploited, is a question the judgment answers for now but cannot guarantee for the future. That the reference has since materialised in Tasleem Ahmed, where a coordinate bench, while making the reference, simultaneously granted the accused interim bail for six months, suggests that the very uncertainty Andrabi sought to resolve may itself have become a ground for enlargement. Andrabi offers a constitutional vocabulary for the argument and, in its factual analysis, a quiet warning that admissibility-blind adjudication is costly, measured in the life years of an incarcerated undertrial on the basis of evidence that is beyond the scope of admissibility of India’s evidence law.
