The IT Act Was Always About Speech, Never About Technology

Read the long title of the Information Technology Act 2000 and you will be told it is a statute for electronic commerce. Read its Chapter XI, headed simply Offences, and you will discover that the same statute carries the most consequential body of speech-policing law that independent India has ever enacted in a single chapter.⁴ Both descriptions are true and both are misleading. The IT Act is a speech statute. It was always a speech statute. The technical framing is the device that allowed it to become one without being treated as one. Shreya Singhal v Union of India struck down one provision and the apparatus carried on regardless.¹ That is the point I want to argue here, because if the framing does not change, neither will the pattern. The settled view is that 66A was an aberration in a technology statute. I think it was the clearest expression of what that statute always was, and that the price of mistaking aberration for type has been a decade in which the speech-control project has advanced under cover of regulatory due diligence. The polemic that follows is against the framing, not against regulation as such.

The settled view, fairly stated

The settled view — held by the bar, the bench, and most of the academic literature — is that the Information Technology Act is technology law that occasionally has speech-law content. On that reading, Section 66A was an aberration: a badly drafted speech-policing provision smuggled into an otherwise sound technology statute, which the Supreme Court excised in Shreya Singhal. The blocking power in Section 69A and the safe-harbour scheme in Section 79 are, on this view, technical regulatory tools. The Intermediary Guidelines, first in 2011 and then in 2021, are framed as due-diligence rules for technology platforms. The Telecommunications Act 2023 is treated as the modernisation of an old telegraph statute. Each piece, taken on its own and against the technical framing it advertises, looks like ordinary regulatory machinery. The settled view is not held in bad faith. It is the natural way to read these statutes if you take their long titles seriously.

I will not pretend the settled view has no force. The IT Act does contain provisions that have nothing to do with speech: rules for digital signatures, electronic records as evidence, certifying authorities, jurisdiction over computer-related offences such as identity theft under Section 66C and cheating by personation under Section 66D. The technical content is real. The settled view rests on it. If the question were only does the statute contain technical content?, the answer would be yes, and the framing debate would be over. But that is not the question that matters. The question that matters is which content does the constitutional work. The settled view's strongest move is to point at Sections 65, 66B, 66C, 66D and say these are not speech provisions; they are property and identity provisions in electronic form, and they are most of what the statute does. The strongest move is also the move that conceals what has changed since 2008. The provisions that have grown — that have been amended, expanded, supplemented by subordinate legislation, and pressed into service by successive governments — are the speech provisions. The property provisions sit, more or less, where the 2008 amendment left them. The speech provisions are the ones in motion. A statute is judged by its centre of gravity, not by the corner where it is calmest.

There is a second prop under the settled view that deserves a fair statement. Speech on the internet is not the same as speech in print or speech at the podium. It moves faster, reaches further, and inflicts harms that the older speech-law architecture was not designed to address — coordinated harassment, deepfakes engineered to deceive, child sexual abuse imagery, organised disinformation campaigns paid for from abroad. A technology framing, the argument runs, is not a dodge; it is an acknowledgement that the medium is genuinely different and that medium-specific regulation is the appropriate constitutional response. The Court in Shreya Singhal itself accepted at paragraph 98 that there is an intelligible differentia between internet speech and other speech for Article 14 purposes — the Article 14 challenge to 66A failed. The settled view leans on that holding. It says: the Court has itself acknowledged that internet speech is different, and a technology framing is the natural way to express that difference. I take this seriously. I will return to it.

Where the settled view fails

The settled view fails at the same place at which the constitutional analysis ought to begin. When a regulatory instrument both (a) authorises the state to control what citizens can say to one another and (b) is justified to the courts as a technology measure, we are in the presence of a category error with consequences. Constitutional review under Article 19(2) demands a tight, narrow reading of every restriction on speech. Constitutional review of a technology statute does not. The framing chooses the standard.

Consider what Section 66A actually was. The text criminalised any information sent through a computer resource or communication device that was grossly offensive, of menacing character, or sent for the purpose of causing annoyance, inconvenience, danger, obstruction, insult, injury, criminal intimidation, enmity, hatred, or ill-will.³ There is nothing technological in that. The medium is electronic; the offence is the speech. The Supreme Court in Shreya Singhal did the obvious thing — applied the standards laid down for free speech in cases stretching from Romesh Thappar v State of Madras through Superintendent, Central Prison v Ram Manohar Lohia and S Rangarajan v P Jagjivan Ram, and found that the section took within its sweep protected speech and was unconstitutionally vague.⁵ What is striking, and what the literature on Shreya Singhal has not made enough of, is that the Court reached this result by reading Section 66A as a speech provision in the first place. The Attorney General's defence, which leaned on the technical medium, was rejected outright: the Court noted that the wider circulation possible over the internet cannot restrict the content of the right under Article 19(1)(a) nor can it justify its denial. The framing was up for grabs and the Court chose the right one. That choice was not preserved. The pattern since Shreya Singhal has been to advance the speech-control project through instruments that present themselves as technology law and that, by virtue of that presentation, escape the standard of scrutiny Shreya Singhal would otherwise demand.

The first instance is the Intermediary Rules 2021.² Part II of these rules imposes due-diligence obligations on social media intermediaries. Most of it reads like compliance regulation: appoint a resident grievance officer, publish a privacy policy, retain records for a defined period. But Rule 4(2) requires significant social media intermediaries providing messaging services to enable identification of the first originator of any information. The official framing is that this is a technical capability requirement — a configuration setting for a class of platforms. Read as a speech rule, it is the architecture of forensic surveillance over private communication. WhatsApp's challenge to that rule in the Delhi High Court rests on the structural impossibility of complying without breaking end-to-end encryption. The case has crawled. Meanwhile the rule remains on the books, untested at the apex. Part III of the same rules creates a three-tier oversight machinery for digital news publishers and OTT platforms, culminating in an Inter-Departmental Committee inside the Ministry of Information and Broadcasting which can order takedown of content. There is no judicial oversight at any tier; the final adjudicator is an executive committee. Stay orders have been granted by the Bombay High Court in Agij Promotion of Nineteenonea Media v Union of India and by the Madras High Court in Digital News Publishers Association v Union of India, but the rules have not been struck down. They sit in a litigation limbo that already favours the regulator, because the regulator does not need a final ruling — it needs the chilling effect, and the chilling effect is operating already.

The second instance is the Fact-Check Unit notification of 2023, which amended Rule 3(1)(b)(v) of the Intermediary Rules to require intermediaries to remove content that a government Fact-Check Unit had flagged as fake, false or misleading about Central Government business. The Bombay High Court struck this down in Kunal Kamra v Union of India, ultimately by a 2-1 split decided by the tie-breaker judgment of Chandurkar J in September 2024.⁶ The doctrinal grounds were unsurprising: vagueness, overbreadth, the absence of any judicial check, the violation of Article 19(1)(a). The grounds are exactly the grounds on which Shreya Singhal struck down 66A. The mechanism is exactly the same: a vaguely defined category of impermissible speech, executive determination, no judicial input. The packaging is different. Section 66A was a penal provision; the Fact-Check Unit rule was a piece of intermediary due-diligence regulation. The substance is identical. The Court correctly saw through the packaging. The government had bet that the technology framing would do the work, and almost won — a 2-1 split is uncomfortably close, and the dissent of Patel J does not read as a fringe view; it reads as a serious judge who took the technology framing on its own terms and reached the result that framing produces.

The third instance is the Telecommunications Act 2023, which I include because the IT Act framework cannot be assessed in isolation from the regulatory machinery that surrounds it. Section 20(2) gives the central government power to take temporary possession of any telecommunication service or network in the interest of public safety or emergency. Section 22 gives the central government the power to make rules on cyber-security, including identification of the user.⁹ The traceability function that Rule 4(2) of the Intermediary Rules sought to lodge inside the IT Act framework has now found a second home inside the telecom statute. If you read the Telecommunications Act as a technology statute, the identification rule is a routine matter of network management. If you read it as a speech statute, the identification rule is the destruction of anonymous communication. Anonymous communication is not a fringe concern; it is the modality through which whistleblowers reach journalists, through which sources reach civil-society researchers, through which dissenters in small towns reach larger publics. The destruction of anonymity is, in substance, the destruction of a category of speech. The technology framing does not make it any less so.

The structural reason the pattern keeps repeating

The pattern repeats because the framing repeats. And the framing repeats because the statute was, from its inception, advertised as something other than what it primarily is. The IT Act 2000 was drafted in the slipstream of the UNCITRAL Model Law on Electronic Commerce. The model law was a piece of commercial legislation for electronic signatures and records. India bolted on a chapter of offences and called the result a technology statute. The Information Technology (Amendment) Act 2008 expanded Chapter XI considerably — introducing Sections 66 through 66F, Section 67A (sexually explicit content in electronic form), Section 67B (child sexual abuse material), Section 69 (interception), Section 69A (blocking), Section 69B (monitoring and collecting traffic data), Section 79 (safe-harbour). Roughly half of the post-amendment Chapter XI is, in substance, speech regulation. None of it is described that way in the parliamentary record. The Standing Committee report on the 2008 Bill devotes its attention to cyber-terrorism, identity theft, and pornography — categories that are real but that obscure the size of the speech-regulatory expansion. The 2008 amendment was passed in the days after 26/11; the parliamentary debate ran to under an hour. There was no separate scrutiny of the speech provisions because there was no recognition that they were speech provisions.

This is the structural point and it is the point on which I disagree most sharply with the settled view. The IT Act is not a technology statute that occasionally regulates speech. It is a speech statute that occasionally regulates technology. The ratio of substantive speech-control content to substantive technical content in Chapter XI alone has been a speech-statute ratio at least since 2008. The framing has not caught up. Lawrence Liang has argued in his chapter on free speech in The Oxford Handbook of the Indian Constitution that Indian free-speech doctrine has always operated within a constitutive split between the universal speaking subject imagined by Article 19(1)(a) and the infantilised, easily provoked subject imagined by Article 19(2).⁸ Article 19(2) was drafted, in part, by people who feared what the colonial subject would say if left unsupervised. The Constitution (First Amendment) Act 1951 widened that supervision in response to the Court's narrow readings in Romesh Thappar and Brij Bhushan, adding public order to the permissible grounds. The IT Act framework, I would argue, is the most recent and most successful expression of the same impulse. It treats the citizen as a source of dangerous speech that has to be managed at the level of the medium, rather than the citizen as a bearer of a fundamental right whose restrictions must satisfy a narrow constitutional test. The technical drag is precisely how the older impulse was given new clothes for the digital era. What Article 19(2) could no longer do directly — because Indian courts had, since the 1950s, tightened the doctrinal vice on direct speech restrictions — was outsourced to a parallel framework whose very name signalled this is something else.

The framing is not a curiosity. It changes which standard of review applies, which precedents the bench reaches for, which arguments counsel build, and which institutions feel that they have standing to object. Treat the Intermediary Rules 2021 as technology regulation and a Standing Committee on Subordinate Legislation reviewing them will ask whether the rules conform to the parent Act; treat them as speech regulation and the same committee will ask whether they pass the proportionality test under KS Puttaswamy v Union of India.¹¹ Treat Section 69 interception powers as technical access controls and the operative test is the procedural template from People's Union for Civil Liberties v Union of India; treat them as speech-and-privacy intrusions and the operative test is whether interception of communicative content can satisfy the three-fold standard of legality, necessity, and proportionality across the wider scale of internet communication.¹⁰ The framing chooses the standard. The standard chooses the outcome.

There is a third effect of the framing that gets little attention. Speech statutes attract a particular form of civil-society scrutiny. They draw the attention of press bodies, of writers' associations, of constitutional-law clinics, of digital-rights groups, of the bar associations. Technology statutes attract a different and narrower set of constituencies: industry bodies, technical-standards organisations, compliance lawyers. When the IT Act is framed as a technology statute, the audience for its scrutiny shrinks. The Internet Freedom Foundation has been a near-solitary public-interest litigant on a number of intermediary-rule challenges. That is a function of the framing as much as of the topic. A speech statute attracts speech-statute opposition. A technology statute attracts technology-statute lobbying. The IT Act framework has been able to grow as quickly as it has because the constituency that would resist it has, for the most part, not been activated. The framing has not just changed the doctrine; it has changed the politics.

The strongest case against this argument, addressed honestly

The strongest opposing case has two legs. The first is medium-specificity, which I sketched in the second section. The second is the institutional argument, and it is the harder one.

The medium-specific argument runs that internet speech is so different in scale and speed from older forms of speech that a separate regulatory architecture is required, and that the technology framing is the honest description of that architecture. The medium is different. The harms are different. A reasonable speech-regulation framework for the internet is going to look different from one designed for newspapers. Anuradha Bhasin v Union of India, in 2020, recognised that internet access itself implicates Article 19(1)(a) and Article 19(1)(g), and required proportionality and reasoned orders for any shutdown.⁷ That is a doctrinal move that responds to medium-specificity without giving up the speech-statute frame, and it is the move I want the rest of the IT Act framework to follow. The defence is honest about a real fact. But the defence proves a more limited point than its proponents need. It establishes that the medium-specific argument has a place in the constitutional analysis. It does not establish that the medium-specific framing is enough to displace the speech-statute framing at the level of statutory characterisation. To say this statute is medium-specific is not the same as saying this statute is not a speech statute. The Code of Criminal Procedure has medium-specific provisions for electronic evidence; it remains a criminal-procedure statute. The Indian Penal Code has been amended to cover offences committed through electronic means; it remains a penal statute. Medium-specificity does not change the centre of gravity. The IT Act's centre of gravity is speech. The medium-specific argument cannot do the work of moving it.

The institutional argument is harder. The case for the technology framing, made by its most thoughtful defenders — lawyers and former officials such as those who staffed MeitY through the drafting of the 2021 Rules — is that the technology framing has produced a coordinated machinery of regulation that no speech-statute framing could have produced. MeitY can talk to MIB which can talk to DoT which can talk to MHA. Rules can be drafted, notified, and amended on a regulatory timescale rather than a legislative one. Platforms can be brought to the table for closed-door consultations of a kind that would be politically impossible if the conversation were openly about speech. The internet has presented governance challenges that the speech-statute framing — slow, court-supervised, every restriction held against Article 19(2) — would have struggled to address. The technology framing is, on this account, what allowed the Indian state to acquire any digital governance capacity at all. Strip it away, and you do not get pristine free-speech jurisprudence; you get regulatory paralysis.

I want to take this seriously because it is not wrong about the empirical record. The technology framing has produced a coordinated machinery of regulation. The speech-statute framing would not have produced the same machinery on the same timescale. The defenders are right about that. They are wrong about what follows. The choice is not between coordinated regulation and regulatory paralysis. It is between coordinated regulation that is open about being speech regulation, and coordinated regulation that pretends otherwise. The first form is harder to build but easier to defend in a constitutional democracy. The second form is easier to build but produces, repeatedly, the kind of provision that Shreya Singhal and Kunal Kamra have to strike down. The pattern of constitutional litigation we have witnessed since 2015 is not the cost of a working system; it is the cost of a framing that has been allowed to run for too long. The institutional argument is best read as a confession: yes, the framing has allowed us to do things we could not otherwise have done. The right response is to ask whether those things should have been done at all, not whether the framing should be preserved so that we can do more of them.

What changes if we accept the speech-statute reading

If the IT Act framework is read as the speech statute it is, three things change. First, the test for every new rule, every new amendment, every notification under Section 69A or Rule 3 of the 2021 Rules, becomes the strict-scrutiny test of Shreya Singhal. Vagueness, overbreadth, and chilling effects are not exotic objections to be raised by digital-rights lawyers — they are the standard objections that any restriction on speech must answer. The burden shifts onto the regulator. Section 69A blocking orders, which today travel as confidential notifications with no public reasoning, would have to be issued with reasons and made reviewable, in the way that any reasoned speech restriction must be. The opacity that currently surrounds blocking — the URL-by-URL list of takedown directions that the public never sees — would have to give way.

Second, the institutional set changes. Speech regulation in India has always operated within a culture of inter-branch suspicion: legislature passes, executive enforces, judiciary reads down. The IT Act framework has, instead, operated within a culture of inter-agency cooperation: MeitY rules, MIB enforces, DoT coordinates, MHA flags. The judicial seat in that arrangement is currently empty. Move the framing and the seat fills. A speech regulator who knows that every action will be scrutinised against Article 19(2) behaves differently from one who does not. The 2021 Rules went through three rounds of consultation that, in retrospect, were not consultation but information-gathering. A speech-statute framing would have demanded a different consultation — open submissions on the impact of each due-diligence obligation on the speech rights of users, not on the compliance costs to platforms.

Third — and this is the consequence I care about most — the political conversation around digital speech changes. The current conversation centres on platforms: what should X do, what should WhatsApp do, what should YouTube do. That conversation is downstream of a prior question that has been treated as settled: what may the government tell the platforms to do? The IT Act framework has answered that question quietly, in language that does not announce itself as a speech-law answer. A speech-statute reading forces the prior question back into view. Once it is back in view, it can be argued. Once it can be argued, it can be lost or won on its merits, which is what a constitutional democracy is supposed to do with questions of this size.

I am not arguing that the internet should be unregulated. The harms are real. Child sexual abuse imagery, non-consensual intimate imagery, coordinated harassment, deepfakes engineered to deceive — the case for some regulatory response is unanswerable. I am not arguing that Section 69A blocking power should be struck down — Shreya Singhal upheld it and was right to do so, subject to procedural safeguards that the 2009 Blocking Rules supply imperfectly. I am not arguing that intermediaries should bear no responsibilities; safe-harbour was always conditional. The argument is narrower and, I think, more important. It is about how we describe what we are doing. Description chooses framing; framing chooses doctrine; doctrine chooses outcomes. The IT Act framework will continue to grow. There will be more rules, more notifications, more amendments. There will be successor statutes — the Digital India Act has been in draft since 2023 and will, when it arrives, replace much of the existing architecture. If it arrives advertised as a technology statute, the cycle resumes. If it arrives advertised as the speech statute that it will, in substance, be, the cycle breaks. Shreya Singhal showed that the cycle can be broken in a single section. It can also be broken at the level of the framing — but only if we are willing to call the statute by its true name. Section 66A is gone. The thinking that produced it is still drafting.

Get the week's biggest decisions, distilled.

A weekly digest for lawyers, students and the legally curious. Every Friday morning. Free.