On September 30, a Delhi High Court bench dismissed the latest bail appeals of Umar Khalid and Sharjeel Imam, the third such rejection for Khalid and the second for Imam. The bench did not weigh the evidence against them afresh. It held, in its own words, that it was “bound by judicial discipline” to follow the Supreme Court’s January 5, 2026 order, under which Khalid and Imam cannot renew their bail applications until the prosecution’s protected witnesses are examined, or until a year has passed—whichever comes first. Both men have now been in custody for more than six years without a concluded trial.

That earlier order is worth returning to, because it did more than deny bail to two men. It granted bail, in the same breath, to five of their co-accused—Gulfisha Fatima, Meeran Haider, Shifa-ur-Rehman, Mohammad Saleem Khan and Shadab Ahmed—all charged in the very same larger-conspiracy case arising from the 2020 Delhi riots. What should trouble anyone reading that order closely is how little it said about why Khalid and Imam were treated differently. It said they stood on a “qualitatively different footing,” citing “a central and formative role” attributed to them. It did not say what conduct made that role central.

The Bail Bar Is High, but Not Absolute

Section 43D(5) of the Unlawful Activities (Prevention) Act (UAPA) is one of the most demanding bail provisions in Indian criminal law. It tells a court it cannot release an accused person if, after looking at the case diary and the police report, it finds “reasonable grounds for believing that the accusation against such person is prima facie true.” For years, courts read this narrowly and defensively—in NIA v. Zahoor Ahmad Shah Watali (2019), the Supreme Court told trial courts to take the prosecution’s material more or less at face value, without weighing it the way a court normally weighs evidence. That reading made UAPA bail close to impossible.

But Watali was never meant to be the last word. A parallel line of reasoning insists that “reasonable grounds” has to mean something more than a bare accusation—it requires what courts have called substantial probable cause, a standard higher than what is needed merely to frame a charge. And in 2021, the Supreme Court in Union of India v. K.A. Najeeb went further still: it held that Section 43D(5) is not an absolute bar. It is, in the Court’s own words, “merely another possible ground” for refusing bail—one that cannot be stretched to justify indefinite custody once a trial has stalled beyond what is reasonable.

That last point is the crux of the Khalid case. The prosecution has had six years to build a case that survives scrutiny. It has not concluded a trial. According to reporting on the proceedings, its own material has already been shown, in parts, to rest on a discredited witness and on an unproven claim about a conspiratorial meeting. If the state’s evidence cannot hold up after six years of opportunity, the “reasonable grounds” finding that Section 43D(5) demands starts to look less like a considered judicial assessment and more like an inherited assumption nobody has gone back to test.

Each Court Defers to the One Above It

The Delhi High Court’s September 30 order does, at least, draw one further distinction. Khalid and Imam had sought parity with two co-accused, Tasleem Ahmed and Khalid Saifi, whom the Supreme Court had granted interim bail. The High Court rejected the claim, holding that Ahmed and Saifi had played different roles in the alleged conspiracy. On its face, this looks like the missing explanation.

Look closer, though, and the explanation does less work than it appears to. The High Court was not independently assessing the evidence against Khalid and Imam; it said plainly that it was bound by judicial discipline to follow the Supreme Court’s order, not free to revisit the underlying material itself. A court that defers to a prior order’s conclusion is not the same as a court that has tested that conclusion. The “different roles” finding, as reported, is asserted rather than demonstrated—it tells us that a distinction exists, not what specific conduct grounds it, nor why that conduct clears the higher evidentiary bar Section 43D(5) is supposed to demand. Six years and three rounds of litigation later, the reasoning offered for continued detention remains conclusory where the statute requires it to be substantiated.

The broader pattern is still the one worth naming: of the accused in this conspiracy case who have sought bail on comparable facts, several have secured it while Khalid and Imam have not, and the courts asked to explain that gap have each, in turn, deferred to the level above them—the High Court to the Supreme Court, the trial court to the High Court—without any one forum fully setting out the evidentiary basis for the distinction in the specific, case-by-case terms Section 43D(5) requires.

Delay Can Itself Become the Punishment

India’s constitutional jurisprudence has answered this kind of question before. In Hussainara Khatoon v. State of Bihar (1979), the Supreme Court held that Article 21’s guarantee of life and personal liberty carries within it a right to a speedy trial — because no procedure that fails to produce a reasonably quick trial can be called fair, reasonable, and just. The Court tied this directly to the presumption of innocence: prolonged undertrial detention, it said, effectively punishes the accused before guilt has ever been established.

That principle has not gone quiet in the years since. In May 2026, the Jammu and Kashmir and Ladakh High Court, ruling in a separate UAPA matter, held that prolonged incarceration combined with trial delay violates Article 21 — and it said so in terms that apply with equal force here: the gravity of an offence, however serious, cannot by itself override constitutional guarantees available to every accused person. Every accused, the Court said, continues to enjoy the presumption of innocence until guilt is established through a fair trial. That is not a radical proposition. It is the ordinary operating principle of criminal justice, restated because it keeps needing restating in UAPA cases.

Set against this backdrop, Khalid and Imam’s continued detention exceeds the benchmarks the Supreme Court itself has treated as significant in comparable cases. Najeeb secured bail after roughly five years. So did others accused alongside him in a related matter. Khalid has now passed six, and the courts asked to revisit that fact keep answering a procedural question—are we bound by the order above us—rather than the constitutional one: has delay itself become the punishment.

A Bail Order Must Say Why

This is not an argument that Umar Khalid is innocent. That is not a question this piece is equipped, or entitled, to answer—that is precisely what a trial is for, and precisely what has not happened. The argument is narrower and, I think, harder to dismiss: a bail provision that requires “reasonable grounds” cannot be satisfied by an assumption that gets passed down from court to court without any one of them actually testing it against the evidence. Constitutional courts owe litigants — all litigants, regardless of the political temperature around their case—a bail order that says why the law’s threshold has been met, not one that cites judicial discipline as a substitute for that explanation. Six years in, deference is not a reason. It is the absence of one.

Sahil Najar is an Advocate enrolled with the Jammu & Kashmir and Ladakh Bar Council, practicing at the District Court, Srinagar. He holds a B.A. LL.B. from the School of Legal Studies, Central University of Kashmir, and writes regularly on Indian constitutional and criminal law for JURIST, LiveLaw, Lawctopus Academike, and Legal Service India.