11.1 Introduction

In April 2023, the Government of India amended a set of regulations called the Information Technology Rules, which primarily dealt with issues around online intermediary liability and safe harbour.Footnote ¹ Until 2023, these rules required online intermediaries to take all reasonable efforts to ensure that ‘fake, false or misleading’ information was not published on their platforms.Footnote ² Previous iterations of these rules had already been challenged before the Indian courts for imposing a disproportionate burden on intermediaries, and having the effect of chilling online speech.Footnote ³ Now, the 2023 Amendment went even further: it introduced an entity called a ‘Fact Check Unit’, to be created by the government. This government-created unit would flag information that – in its view – was ‘fake, false or misleading’ with respect to ‘the business of the central government’.Footnote ⁴ Online intermediaries were then obligated to make reasonable efforts to ensure that any such flagged information would not be on their platforms. In practical terms, what this meant was that if intermediaries did not take down flagged speech, they risked losing their safe harbour (guaranteed under the Information Technology ActFootnote ⁵).

The 2023 Amendment was immediately challenged before the High Court of Bombay, and at the time of writing, hearings were in progress. A perusal of the pleadings in the case reveals that the state’s defence of the 2023 Amendment rested on a few important aspects.Footnote ⁶ First, the state highlighted the dangers of online ‘fake news’, its threat to public order, and that the ‘virality’ of the Internet made counter-speech an unviable solution. Second, as a matter of constitutional law, the state argued that ‘fake’ or ‘false’ news was unprotected by the constitutional guarantee of freedom of speech and expression.Footnote ⁷ Citing US Supreme Court judgments such as Hustler Magazine v. Falwell, it noted that fake news had no ‘constitutional value’, and therefore did not deserve constitutional protection. And third, it argued that the spread of fake news undermined other constitutional rights, such as the right to informationFootnote ⁸ and – by extension – the right to vote (which the state read as a right to vote based on true and accurate information). This, in turn, allowed the regulation of online fake news in order to preserve other equally important constitutional rights.

The state’s arguments in the IT Rules challenge reveal an inherent tension within Indian free speech jurisprudence. In this chapter, and keeping the IT Rules challenge as the background framework, I will excavate this tension, and tease out its implications for the regulation of misinformation and disinformation in the contemporary context. Broadly, this tension is manifested in two lines of jurisprudence. The first line can be broadly characterised as the ‘autonomy’ approach to free speech. Classically liberal in orientation – and drawing from US First Amendment jurisprudence, although not as expansive as in the USA – the ‘autonomy’ approach advocates for a degree of content-neutrality in restrictions to free speech. Its underlying principle is that autonomous individuals decide for themselves how to receive and respond to speech. The second line can be called the ‘perfectionist’ approach to free speech (I use ‘perfectionist’ in the sense that it is used by political philosophers, such as Joseph RazFootnote ⁹). The perfectionist approach as it appears in Indian jurisprudence eschews content neutrality, links the degree and nature of constitutional protection that speech has to its ‘value’, and is comfortable with the state being the arbiter of value. While these two lines of jurisprudence have proceeded on an almost parallel track in the seven decades of constitutional history, it is in cases such as the IT Rules challenge – I will argue – that they come to a head; and it is in contexts such as that of contemporary online disinformation that both lines reveal their limitations, and demonstrate the need for a new way of thinking about the relationship between speech, misinformation and the Constitution.

11.2 The Autonomy Line

Article 19(1)(a) of the Indian Constitution guarantees to all citizens the right to freedom of speech and expression. Article 19(2) authorises the state to make, by law, ‘reasonable restrictions’ upon the exercise of this freedom, ‘in the interests’ of eight different categories. These include, for example, ‘public order’, ‘defamation’, ‘incitement to an offence’ and so on.Footnote ¹⁰ The structure of India’s free speech clause largely resembles the familiar, two-tiered structures that we see in many constitutions: the declaration of the right, followed by an exhaustive list of circumstances under which it may be restricted.Footnote ¹¹ This approach precludes ‘content neutrality’ in the US First Amendment sense, as Article 19(2) itself authorises the restriction of speech on the basis of its content. The proof of the pudding, however, is in the interpretation.

The debates of the Constituent Assembly reveal the tension underlying the free speech clause right from the outset: liberal arguments in favour of minimising the restrictions clause were met with perfectionist arguments extoling the expertise and legitimacy of the newly elected government.Footnote ¹² This internal tension has then spilled over into the judicial interpretation of the free speech clause.

Right from the beginning, a set of Indian judgments cited classic liberal arguments – from John Stuart Mill to Oliver Wendell Holmes’ ‘marketplace of ideas’ – as the underlying foundation of the free speech clause. This is most clearly visible in the evolution of Supreme Court jurisprudence on socially subversive, or dissident, speech. In 1960, the Supreme Court set aside the criminal conviction of a politician who had been prosecuted for exhorting citizens not to pay a tax that he considered unjust.Footnote ¹³ Responding to the state’s arguments that incitement to disobedience of a law could spark a general revolution, the Court insisted on a ‘proximate link’ between speech and the feared public disorder before it could be restricted or criminalised.Footnote ¹⁴ A few years later, the Court used the colourful analogy of a ‘spark in a powder keg’ to describe the degree of proximity that was required between speech and public disorder.Footnote ¹⁵ Most recently, while considering certain broadly worded provisions of the Information Technology Act, the Court drew a distinction between ‘advocacy’ and ‘incitement’, holding that only the latter could be criminalised consistent with the free speech clause of the Constitution.Footnote ¹⁶

The distinction between ‘advocacy’ and ‘incitement’, in particular, is a clear statement of the autonomy approach to free speech. It brings to mind Justice Brandeis’ famous observation that as long as there is time to respond, the remedy to subversive speech is counter-speech, and not censorship.Footnote ¹⁷ The idea underlying this is that the default presumption is that the autonomous listener is responsible for choosing how to take – and respond – to speech, except in narrowly defined cases of diminished autonomy.Footnote ¹⁸ The classic example given is of shouting ‘fire’ in a crowded theatre, a situation where the listeners have no time to evaluate the truth of the speech but must act immediately. The requirement of imminence – the Brandenburg test, as it is known under First Amendment jurisprudenceFootnote ¹⁹ – broadly tracks this underlying theory, limiting itself to situations (for example) such as inciting an already enraged mob to commit acts of direct violence.Footnote ²⁰

The autonomy approach to free speech is also evident in other aspects of Indian free speech jurisprudence: in particular, in cases dealing with the question of whether some forms of speech are inherently more ‘valuable’ than others. In a series of early cases, for example, the Supreme Court beat back the government’s attempts to curtail the volume of advertisements in newspapers by drawing a direct link between advertisements, income flow, volume of circulation and – thereby – the newspapers’ right to free speech.Footnote ²¹ Another example is that of the ‘commercial speech doctrine’ issue, where the Court departed from earlier jurisprudence to hold that commercial speech was as much protected by Article 19(1)(a) as any other form of speech.Footnote ²²

This vein of Supreme Court jurisprudence is what the petitioners in the IT Rules case tapped into in order to rebut the state’s argument that ‘fake’ or ‘false’ speech is excluded from Article 19(1)(a) protection altogether, because it possesses low or no value.Footnote ²³ The autonomy approach to free speech – which regards listeners as autonomous individuals, responsible for their own choices – precludes a priori assessments of value, and any attempt to link constitutional protection to the value of speech, other than that which is already in the Constitution. The basic idea is that judgements of value are to be made by the listeners and not by the state. The structure of Article 19(1)(a) and (2), which specifically sets out the kinds of content that can be regulated, appears to support arguments against additional value judgements being made by the state: after all, the argument goes, those value judgements have already been made – exhaustively – within the Constitution itself. And false speech – for both principled and pragmatic reasons – is not one of the categories under Article 19(2).

11.3 The Perfectionist Line

However, even as early Supreme Court judgments were entrenching the autonomy approach to free speech, there were other judgments that took another line, and accentuated the tension inherent within the interpretation of Articles 19(1)(a) and 19(2). Early on, when considering a ban on misleading medicinal advertisements under a law evocatively titled the Drugs and Magical Remedies Act, the Supreme Court upheld the ban by noting that advertisements for magical remedies were unprotected by the free speech guarantee because they made no contribution to the exchange of ideas – political, economic or cultural – that constituted the basis for why there existed a free speech guarantee in the first place.Footnote ²⁴ Here, the Supreme Court constructed a hierarchy between kinds of speech, assigning them relative value based on their ‘contribution’ to the democratic public sphere, and linking the degree or nature of constitutional protection with the value they had. Other than this, an unarticulated major premise of the judgment, of course, was that the target audience of misleading information would not be in a position to sift the wheat from the chaff, or be able to identify the misleading content in medicinal advertisementsFootnote ²⁵ – something that justified state intervention in this domain.

Similar reasoning can be glimpsed in the evolution of the Court’s jurisprudence on obscenity. In its early days, the Court adopted the Victorian British ‘Hicklin test’, which asked whether speech had the tendency to ‘deprave or corrupt’ the minds of those into whose hands it might fall.Footnote ²⁶ Over the course of time, this test was gradually liberalised, until in 2011, the Court replaced it with the US Miller test, where the focus is on whether the offending work appeals solely to the ‘prurient interest’.Footnote ²⁷ However, whether the more strait-laced Hicklin or the more liberal Miller, what unites them both is the belief that certain kinds of speech are inherently valueless. This is made clear by the fact that both judgments – as well as the obscenity provision in India’s Penal CodeFootnote ²⁸ – have exceptions for works with genuine ‘literary’ or ‘cultural’ value.Footnote ²⁹ Once again, therefore, in the realms of obscenity jurisprudence, we have a judgement – made by the state, and subject to judicial review by the Courts – about the value of forms of speech, and linking up perceived value with constitutional protection.

The perfectionist impulse is most strikingly visible in a judgment called Union of India v. Motion Picture Association, where the Court was considering a state requirement for cinemas to show short ‘educational’ or ‘cultural’ documentaries before the start of the film.Footnote ³⁰ This was challenged on grounds of being akin to a ‘must-carry’ provision, tantamount to compelled speech. The Supreme Court’s response is instructive, and merits quoting in some detail:

And:

The excerpted passages are important because they go to the heart of the perfectionist approach to free speech. The Court here drew a clear-cut distinction between ‘propaganda’ or ‘partisan speech’ or a ‘distorted point of view’ and (presumably accurate) information that ‘furthered informed decision-making’. A must-carry rule that required compulsory broadcasting of the former would violate the right to freedom of speech and expression, but a requirement of carrying the latter would not, since its net effect was to further the right by providing information to all, rather than curtailing it.

It is particularly instructive that the Court drew a link between consumer warnings on cigarette packs (a classic example, as indicated above, in the consumer fraud domain, where concerns of autonomy are thought to be of relatively low intensity), with the domain of political speech, where the concerns of autonomy are believed to be at their highest. The overarching theme, then, cuts across domains of speech: the perfectionist approach to free speech, which sees the role of free speech in a democracy as performing a very specific function, and accords constitutional protection to only that kind of speech that is consistent with this function. This necessarily requires the state – and the courts – to construct an internal content-based hierarchy within speech, and to assign social value to forms or kinds of speech.

This, it must be noted, is distinct from a free speech jurisprudence where regulations or restrictions are justified not on the basis of the value of the speech in question but by virtue of its impact on other constitutional principles. Hate speech jurisprudence, for example, is based not on the premise that hate speech is low or no-value speech, but that it is destructive of the constitutional principles of equality and equal protection.Footnote ³³ This jurisprudence is also at odds with a strict articulation of the autonomy approach to free speech.Footnote ³⁴ But at the same time, liberal democracies have been broadly able to reconcile the existence of hate speech laws with a general autonomy-based approach to free speech: there is an understanding that the principles underlying the free speech guarantee need to be balanced against the maintenance of other constitutional principles, which – in turn – will require overriding autonomy in select domains, without conceding its own importance as a constitutional principle.Footnote ³⁵

What is crucial to note is that such arguments do not depend – as already pointed out – on assigning values to forms of speech. The criminalisation or penalisation of ‘misinformation’ or ‘disinformation’ simpliciter, however, raises different concerns. For instance, in the IT Rules case, had the clause referred to ‘fake or false speech’ that ‘incites public disorder’, or ‘fake or false speech’ that ‘vilifies a community of people’, different constitutional questions would have arisen. By contrast, penalising misinformation or disinformation simpliciter is – in effect – penalising bare falsehood. To defend such a provision – which, on its face, falls outside the eight sub-clauses of Article 19(2) – would require the state to construct an argument along the lines of low- and high-value speech, as the only other option is to argue that false speech falls out of the ambit of the free speech clause altogether. As we have seen, this is indeed what the state did, and how it chose to defend the constitutionality of the 2023 IT Rules Amendment when it was challenged in court.

And it is this defence that lays bare the tension between the autonomy and perfectionist approaches to free speech, which have pulled Indian free speech jurisprudence in different directions from the inception of the Indian Constitution. As we have seen, it is in the context of contemporary online misinformation and disinformation – with its virality, and its admittedly corrosive effects on democracy – that these tensions become particularly stark. And in the final section, we shall examine how these tensions play out in this context.

11.4 Misinformation and Disinformation in Contemporary Contexts

As pointed out, the state’s core argument in the IT Rules case was that outright falsehoods are unprotected by the constitutional guarantee of free speech, as false speech has no constitutional value. On the flip side, the state argued that citizens have a right to receive true and accurate information, a right that is frustrated in an online environment awash with falsehoods. In particular, the state linked this right to the right to vote: it argued that false or fake news distorted the voting process – and thereby democracy – by distorting the base of information upon which voters made their decisions.Footnote ³⁶ (This last argument, one will notice, specifically takes issue with the autonomy approach to free speech.) And finally, it flagged the specific character of the Internet and online speech: in particular, virality, which makes it difficult for false speech to be ‘countered’ with facts.

Taken on their own, these arguments do not necessarily sound unreasonable. The constitutional issue, however, lies in the regulatory mechanism that the state devised to deal with the problem: a regulatory mechanism that involved a government-mandated fact-check unit ‘checking’ speech about the government, and backing it up with the coercive consequences of losing safe harbour. To elaborate, safe harbour – recognised by Section 79 of the Indian Information Technology Act – refers to the legal position where intermediaries cannot be held liable for the content that is posted or disseminated on their platforms. Safe harbour is normally a qualified right, imposing certain obligations of due diligence upon the intermediary. In India, this takes the form of ‘actual knowledge’ – that is, the intermediary needs to be notified, either by a court order or a government notice, that it is hosting unlawful content. If, on receiving such knowledge, the intermediary refuses to act, it then ‘loses’ safe harbour, and is open to lawsuits seeking to hold it liable of the offending speech.

Deprivation of safe harbour is therefore not direct censorship, such as a book ban or the blocking of a website. Theoretically, an intermediary can disagree with the government on the issue of whether flagged content is actually lawful or unlawful, defend its assessment in court, and even possibly win. Practically, however, safe harbour is a vital safeguard that allows intermediaries to perform their functions without the constant threat of multiple lawsuits hanging over their heads. An intermediary will – in its own interests – prefer to comply with a government directive rather than lose safe harbour. The coercion is therefore indirect – a threat of the loss of safe harbour and the consequences that follow – and the free speech implications lie in accompanying self-censorship and the chilling effect.

This detour is important, as it was the basis of the petitioners’ argument that the language of the 2023 IT Amendment creates a slippage between regulating misinformation and making the state the arbiter of truth. The possibility of such a slippage is always present in state efforts to tackle misinformation, and – the petitioners argued – the 2023 Amendment reveals how it actually happens. If the consequence of an intermediary not complying with the government Fact Check Unit is the loss of safe harbour, then – as far as the intermediaries are concerned (and by extension, the users of the platform, both the speaker and the recipient of the speech) – the government Fact Check Unit’s assessment of what is true or false is effectively determinative.

This leads to a few problems. The most obvious one, of course, is whether a free speech guarantee is consistent with allowing the state to act as an effective arbiter of truth, even if one were to accept the proposition that false speech is outside the pale of constitutional protection. It is, after all, one thing to say that false speech should be regulated, and another thing to say that the state (and in this case, specifically, the political executive) should be the body that has control over regulation. While the perfectionist line of free speech jurisprudence in India might well support the first proposition, the second presents a greater challenge. The autonomy approach, on the other hand, would reject out of hand the regulation of bare falsehood altogether, as – in the absence of any legally cognisable harm traceable to Article 19(2) of the Constitution – its rationale can be nothing other than an internal hierarchy within Article 19(1)(a).

Secondly – as petitioners’ counsel pointed out – there is a range of statements and forms of speech where the question is not merely whether something is true or false, but whether it falls within the true/false binary at all. For example, consider the judgment of the South African Constitutional Court in DA v. ANC. The case involved a bulk SMS sent by the main opposition party, the Democratic Alliance, which stated that an impartial report ‘shows how [President] Zuma stole your money’ to build his home, whereas the report had not made any finding of ‘theft’ against President Zuma.Footnote ³⁷ Justices of the constitutional court split over the exact meaning of the phrases ‘shows how’ and ‘stole’, with some judges holding that it was an allegation of theft (and therefore, false), while others held that it was an accusation of unethical behaviour (and therefore neither true nor false, only convincing or unconvincing). As the case reveals, therefore, there is a subjectivity underlying not merely whether a claim is true or false, but the nature of the claim itself.Footnote ³⁸

Although this issue would dog any regulatory attempt to curtail misinformation or disinformation online, it acquired a particular salience in the IT Rules case, as the regulatory form was one that proposed to give the executive the effectively determinative power of settling such questions. But although the issues and arguments arose in a specific context, it is evident that they need to be considered by regulators grappling with these questions across the board.

The IT Rules case is complicated by the tensions between the autonomy approach and the perfectionist approach within Indian free speech jurisprudence (by now, it should be obvious that this tension exists in the constitutional jurisprudence of most liberal democracies). It does appear – for the reasons pointed out above – that the IT Rules, in their specific form, go even beyond what is permitted by the perfectionist approach.

At the same time, it also appears that a strict application of the autonomy approach is not entirely appropriate to the era of mass online misinformation, with its well-known corrosive effects on democracy. As the scholarship around hate speech has taught us, it is not only in the Brandeis-type situations of incitement, or shouting ‘fire’ in a crowded theatre, that the remedy of counter-speech fails. There are other reasons – and other contexts in which – counter-speech can fail. This is especially true when one accounts for – in the words of Jack Balkin – the ‘infrastructure’ of free speech, and recognises the fact of unequal access to infrastructure determining who can speak, how much and to what extent.Footnote ³⁹ The question of virality on the Internet – and specifically virality of misinformation on the Internet – often tracks inequalities of power and resources in the offline world, which are transplanted into the online domain.

Thus, neither the autonomy approach nor the perfectionist approach – the two strands of Indian free speech jurisprudence – might give us all the adjudicatory resources we need to fit the regulation of online disinformation and misinformation within the contours of constitutional jurisprudence. However, there are other, promising avenues: for example, in a one-off judgment, the Supreme Court has recognised a right of reply as forming a part of the right to free speech;Footnote ⁴⁰ another has recognised the importance of access to the infrastructure of speech.Footnote ⁴¹ How this might work in the online domain remains an open question – for instance, the success of Twitter (now X) ‘Community Notes’ – as a community-moderated reply – is much debated. What is clear, however, is that any attempt must draw both from the autonomy-respecting strand of Indian free speech jurisprudence, while also taking seriously the problem of online misinformation, in shaping the nature and the form of regulatory remedies. The IT Rules perhaps are an example of how not to do it: that is, by eschewing autonomy entirely, taking the perfectionist approach to its extreme, and opting to combat misinformation by making the state the arbiter of what is misinformation and what is not.