[This is a guest post by Tanmay Durani.]
The Delhi High Court’s judgment in Telegram FZ LLC v Union of India (W.P.(C) 8259/2026), delivered by Justice Tejas Karia on 19 June 2026, concerns the temporary nationwide blocking of an entire messaging platform under Section 69A of the Information Technology Act, 2000, in response to fraud surrounding the NEET-UG re-examination. At one level, this is simply another examination malpractice case, the sort of dispute that appears with some regularity in Indian administrative law. The more consequential aspect of the judgment, however, lies elsewhere. To my knowledge, it is the first time a court has more or less accepted that Section 69A, which was enacted to block specific information, may extend to blocking an entire platform when the State argues that the platform’s design prevents unlawful content from being separated out from lawful content. The question this raises is not whether examination fraud is a serious problem, which it is. The question is whether the structure of Section 69A, as the Supreme Court understood it in Shreya Singhal v Union of India, can sustain the extended interpretation of “information”, and whether the proportionality analysis that is supposed to condition such an interpretation was, in this case, actually performed.
The Court’s Two Moves
The judgment proceeds along two analytically distinct tracks, though the Court does not always keep them apart. The first concerns administrative law: was the Secretary’s satisfaction under Rule 9 of the 2009 Blocking Rules an independent one, or merely reiterating the requesting agencies’ case? The second concerns constitutional and statutory interpretation: does the blocking power in Section 69A extend to an entire intermediary, and if so, on what terms does the doctrine of proportionality permit its exercise?
On the first question, the Court may well be correct, although its reasoning leaves something to be desired. Telegram’s objection was that the 16 June 2026 Order mostly tracks the language of Section 69A and reproduces the requesting agencies’ account of events, without revealing any distinct evaluative exercise undertaken by the Secretary himself. Seen through an administrative-law lens, this raises a straightforward issue of non-application of mind. There is an important difference between recording satisfaction and genuinely reaching it. Where Parliament or the legislature requires an authority to be satisfied about certain facts before exercising power, courts have been willing to test whether that satisfaction emerged from the authority’s own consideration of relevant material, rather than from a formulaic adoption of someone else’s view (Barium Chemicals Ltd. v Company Law Board; see also Rohtas Industries Ltd. v S.D. Agarwal).
The Court’s answer (at paragraphs 25 and 28) is that the two-stage structure of the Blocking Rules, an emergency interim direction under Rule 9, followed by a hearing and a reasoned Final Order under Rule 7 – means that the adequacy of reasons must be assessed across the process, not frozen at the moment of the interim order. That makes sense in terms of how the Rules are set up. Rule 9 is there because, in urgent situations, the government is not expected to provide the kind of detailed justification that would ordinarily accompany a blocking order. Requiring that level of reasoning at the interim stage would largely defeat the point of having an emergency mechanism in the first place. The Final Order of 18 June 2026 is, on its face, far more elaborate, engaging Telegram’s submissions point by point.
It’s worth being precise about what is conceded here, because it isn’t much. Accepting that detailed reasons can come after the hearing says nothing about whether those reasons, once given, are the product of actual adjudication or simply echo one party’s version of the facts. The Committee’s conclusion that “granular filtration is technically impossible at this scale” (Final Order, para 5) is treated as a fact already established, yet this was the precise point Telegram disputed, in its Reply and then again in clarifications filed after the Meeting. The Court doesn’t really probe how the Committee approached and reconciled this technical disagreement between Telegram and the Government; all it says is that Telegram’s submissions were heard and ultimately rejected (paras 27-30). That gives almost no basis to judge whether the Committee independently weighed the competing technical claims or essentially deferred to the Government’s account of Telegram’s limitations. Having reasons and having arrived at them independently are not the same thing, and neither is the same as the interim-versus-final distinction the judgment leans on. So the Court’s answer to the timing objection simply doesn’t reach this deeper question.
The Expansion of Section 69A
The second move is the more consequential one, and it is here that the judgment departs furthest from existing scheme of law. Section 69A authorises the blocking of “any information generated, transmitted, received, stored or hosted in any computer resource.” “Information” is defined in Section 2(1)(v) to include, among other things, codes, computer programmes, software, and databases. The Court reasons (paragraphs 35 to 37) that since an application is, in the most literal sense, a compilation of code, software, and databases, an application is therefore “information” within the meaning of the section, and Section 69A consequently authorises the blocking of the application as a whole.
One need not quarrel with the language of Section 2(1)(v) to see the difficulty here. The interpretation has no obvious stopping point. A messaging platform is software, certainly, but so is an email client, a banking application, or a hospital database. If that is enough to trigger Section 69A, then the provision ceases to be about blocking particular unlawful content and starts becoming a tool for controlling the software systems through which information is exchanged. The Court does not pause to ask whether Parliament, in 2008, legislating in the aftermath of 26/11 and primarily contemplating the blocking of specific websites, URLs, and content items, intended “information” to bear this platform-level meaning. Nor does it engage with the fact that the statute keeps blocking under Section 69A structurally apart from intermediary regulation under Section 79 and the 2021 Rules — a separation this blog discussed before, in writing on the Sahyog portal litigation. There, the Karnataka High Court let Rule 3(1)(d) takedown powers substitute for what Section 69A was supposed to handle, and we criticised that for muddying a boundary Parliament had drawn with some care. The Telegram ruling muddies the same boundary, approaching from the other direction entirely. Where Sahyog dressed up intermediary powers as blocking powers, Telegram does something blunter: the blocking power simply swallows up the intermediary.
Much turns on what exactly the Supreme Court thought it was approving in Shreya Singhal. The judgment repeatedly describes Section 69A as narrow, and the reason for that description becomes apparent when one reads the Court’s discussion of the Rules. The process assumes a fairly simple model: particular information is identified, a request is made, relevant parties can be heard, reasons are recorded, and the decision is tested against Article 19(2). The safeguards are calibrated to that model. A platform-level blocking order departs from it in a fundamental way. The target is no longer a discrete item of information but the platform itself, carrying an ever-changing mix of lawful and unlawful content. There is no obvious originator, no singular piece of hosted material, and no straightforward way of matching the order to a particular constitutional ground. Yet the same procedural safeguards are treated as sufficient. The effect is to stretch Section 69A beyond the setting in which its constitutionality was defended, while leaving untouched the safeguards that were designed for a much smaller exercise of power.
Proportionality and Architecture
The judgment’s proportionality analysis (paragraphs 38 to 47) recites the four-part structure from Anuradha Bhasin v Union of India: legitimate aim, rational nexus, necessity, and least restrictive means — and concludes that each is satisfied. What is distinctive about the analysis, however, is not its structure but its object. The Government’s argument rather than turning principally on the content circulating on Telegram, turned on Telegram’s architecture: its cloud-based infrastructure, its bot ecosystem, its capacity to spin up mirror channels the instant an offending one is taken down, and its message-editing function, which the Committee found capable of being used to backdate fabricated “leak” evidence. The Court accepts (paragraph 42 onward) that these features render content-specific enforcement futile, and that this futility is itself what justifies the platform-wide remedy.
This is a genuinely novel form of reasoning, and it deserves to be named as such. Proportionality analysis ordinarily asks whether this restriction is necessary to address this harm, measured against the availability of narrower alternatives. What the judgment does instead is treat the platform’s design — its resistance to narrower enforcement — as itself supplying the proportionality of the broader remedy. The less amenable a platform is to surgical intervention, the more proportionate total intervention becomes. Taking this logic to its end-point, the more effectively a platform is built — the more resilient, decentralised, and resistant to selective takedown — the more vulnerable it becomes to being blocked in its entirety, because resilience is taken to signify unaccountability. A feature engineered for legitimate purposes (resistance to censorship, privacy through pseudonymous handles, resilience against targeted takedown) becomes, on this reasoning, evidence supporting the case for a remedy that dispenses with narrow-tailoring altogether.
The existence of fraud was not at issue in this case. The authorities relied on a substantial body of material: I4C’s findings, multiple arrests, a registered FIR, and even statements by Durov himself acknowledging the problem of “backdating scams” (Para 10–11, Final Order; Para 26 of the judgment). The harder question was whether a complete block was proportionate. Anuradha Bhasin requires courts to examine whether less intrusive alternatives remained available. Yet it is not entirely clear that such an exercise was undertaken here. What the judgment shows is that the executive claimed to have exhausted other options. What it does not show with the same clarity is why those options were insufficient. Paragraph 45’s finding that “entity-specific interventions… were repeatedly found to be ineffective and inadequate” rests on NTA’s account of channels reappearing under new names, which is one party’s characterisation of a fact Telegram disputed, having claimed 900 of 1,300 flagged URLs were already disabled. Anuradha Bhasin required the authority to show it had genuinely canvassed intermediate options — say, geofenced restriction, time-bound holds on forwarding, feature-specific suspension short of a full block — before reaching for the most restrictive tool available. Further, the Judgment seems to be over-relying on the fact that channels re-emerged, but there is little quantification of how successful or unsuccessful the targeted approach actually was, for the court to decide that this approach had failed. A court could have required a more systematic showing, some empirical data that such harm was being caused without stoppage, so the government had no choice but to ban it (see Internet Mobile Association of India vs Reserve Bank of India)
The judgment is rich in evidence that the harm was real, but considerably less rich in evidence that this particular step, that nothing short of a total block could have stopped it, was independently tested rather than adopted from the agency seeking the block.
The Problem of Architectural Liability
The deepest difficulty with the judgment, however, lies a level below its application of proportionality. It lies in the principle the judgment installs for future cases: that the difficulty of enforcing the law against specific bad actors on a platform can itself justify removing the platform from the public altogether. Constitutional law traditionally addresses unlawful conduct — actual speech, actual fraud, actual incitement — rather than around the abstract possibility that a communications technology might be misused. This distinction is embedded in free-speech doctrine itself. In Shreya Singhal, the Supreme Court made clear that public-order restrictions demand a close nexus between speech and the anticipated disorder. The Court was not concerned with expression merely because it created an environment in which disorder might become possible — something akin to the State’s argument here that the platform’s architecture facilitates such outcomes — but with expression directly connected to the disorder itself.
The architectural-liability logic in Telegram threatens to dissolve that distinction. Once the basis for blocking shifts from particular content to the design of a service itself, the implications spread much further. Features such as encryption, disappearing messages, pseudonymous accounts, or limits on centralised moderation are hardly unique to one platform. They exist across a range of digital services. If those features become grounds for state action because they complicate investigation or enforcement, then the same argument could just as easily be directed at encrypted email providers, VPNs, peer-to-peer networks, and other privacy-enhancing technologies. Yet these are technologies that have long been defended as important safeguards for free speech and privacy precisely because they reduce the state’s ability to monitor and control communications. A rule that converts resistance-to-takedown into a ground for platform-wide blocking is, in substance, a rule that prohibits privacy-by-design. The Attorney-General’s submission that a platform “structurally incapable of ensuring accountability” forfeits the right to invoke proportionality (paragraph 15) should be read with this consequence squarely in view: it is an idea that, if generalised, asks not what a platform’s users did, but what a platform’s engineers built, and treats the latter as sufficient grounds for excluding the former from the public sphere.
Conclusion
The facts of Telegram are dramatic enough to attract attention, but they are not what makes the case important. Its lasting significance lies in how it reimagines Section 69A. The judgment moves from blocking particular information to accepting that a platform itself may be the object of a blocking order, and it does so through an expansive reading of the statutory definition of “information.” More strikingly, Telegram’s resistance to targeted intervention becomes part of the justification for shutting it down altogether. If that approach holds, the constitutional debate after Shreya Singhal changes considerably. The question, more than content, now, is also about whether entire digital spaces may disappear from public use because the State finds them difficult to police in narrower ways.
