Two minds sit inside the Supreme Court’s hate speech judgment of April 29, and they do not speak to each other.

The first mind, distributed across Part I of the IV parts of the judgment delivered by the bench of Justices Vikram Nath and Sandeep Mehta, is verbose on fraternity. Spread over 85 of the 125 pages, hate speech is rightly described by Justice Nath, who authored the judgment, as “fundamentally antithetical to the constitutional value of fraternity” which “strikes at the moral fabric of our Republic”. It condemns the “us versus them” mentality and reminds public figures that words have consequences in a diverse society.

The vocabulary is, unsurprisingly, unimpeachable. This is the version of the judgment that will be quoted in seminars and op-ed pages, the version that lets the judge here appear to take hate speech seriously.

The second mind, lodged in Part II and III of the judgment, deals with what hate speech looks like when actual ruling-party politicians and those supporting them speak it. This is where the court got a chance to strictly enforce what it so eloquently pontificated in the previous 85 pages.

Two appeals were before the bench in Part II. The first was a criminal appeal arising from CPI(M) leader Brinda Karat’s complaint against the January 2020 speeches by BJP leaders Anurag Thakur and Parvesh Verma—particularly the infamous slogan “Desh ke gaddaron ko, goli maroon saalon ko”—in the immediate context of the anti-CAA mobilisation. The second was a civil appeal from Telangana concerning Islamophobic posts and hashtags circulated online, and sought criminal action against offending parties.

In the criminal appeal, the Supreme Court first corrected an important legal error: the Delhi High Court had wrongly held that prior sanction was required before a Magistrate could order FIR registration under Section 156(3) CrPC—a welcome decision since such prior sanction was necessary only at the stage of taking cognisance, not for setting the criminal law in motion via FIR registration.

But what about the hate speeches already uttered?

Having corrected the law, Justice Nath’s bench proceeded to judge the merits of the case. In just three short paragraphs, the bench agreed with the Delhi High Court’s conclusion that no offence was made out in the BJP leaders’ case.

The Delhi High Court had relied on a police status report which held that “gaddar” meant “traitor” generically and therefore did not refer to any specific community, and that the second speech was merely about the Shaheen Bagh protest. The Supreme Court did not simply defer to that view procedurally but also independently endorsed the conclusion that no cognisable offence was made out.

But what about the method which is the main problem? Communal political speech in India, particularly since 2014, works overwhelmingly through coded reference rather than explicit naming, called “dogwhistling”. The identity of the “gaddar” addressed at the height of the anti-CAA mobilisation, in the immediate vicinity of Shaheen Bagh, was of course not legally ambiguous but quite politically obvious.

Justice Nath proceeds as if the meaning of words can be assessed independently of audience, context, and purpose. Can one even assume that the judge knew nothing of the term dogwhistling, particularly when pointed out by several counsels of the petitioners? Such wilful naivete fools none.

The dangerous precedent that the Supreme Court has now endorsed is that coded incitement through dogwhistling is immunised from criminal action.

The bench’s correction on sanction is immediately drained of practical effect by endorsing the merits conclusion that the speeches before the bench disclosed no offence. Such a flip-flop stand taken by the court is reminiscent of the Chandrachud court. In that sense, the Supreme Court has completely sidestepped constitutional accountability.

The Telangana appeal received even shorter treatment. In just a single paragraph, the grievance, the bench held, had been “adequately addressed” and no further adjudication was warranted.

This is the operative voice.

It is the one police and magistrates will reach for in the future. And it tells them, in effect, that as long as inflammatory political speech avoids naming a community in unmistakable terms, as long as the politician opts for “gaddar” instead of a religious label, or wraps incitement in deniable references to a protest site, a street, a particular kind of name, the laws addressing hate speech that apparently exists today does not reach them.

In Part I, the bench framed four questions which it answered. Can the court create or expand criminal offences? No. Does the existing legal framework adequately cover hate speech? Yes. Are procedural remedies against non-registration of FIRs, like through approaching a senior police officer or the magistrate if police refuses, sufficient? Yes.

The fourth, and the most important, was the question of judicial monitoring of the law’s enforcement. Should continuing mandamus be issued against the executive? The court said that “in the absence of any legislative vacuum or systemic failure of such magnitude as would warrant continuous judicial monitoring, such a course would be neither justified nor consistent with the principle of judicial restraint”. In other words, Justice Nath’s bench felt that there is no large systemic failure of anti-hate speech law enforcement that would justify its continuing intervention. Judicial restraint is therefore necessitated.

But it is impossible to miss out on the cost of this judicial restraint being borne unevenly.

The court tells petitioners to trust the police station. Then the Superintendent of Police. Then the Magistrate under Section 156(3). Only after that, the writ jurisdiction of the higher courts. As an account of ordinary criminal procedure, this is unobjectionable. But hate speech complaints against politically powerful speakers are not ordinary. The petitioners’ submissions, recorded in the judgment itself, repeatedly alleged that police stand inactive while speeches are delivered in their presence, that FIRs are not registered, that political alignment dictates whose words are investigated and whose are not. Not to forget, the risk of grave personal harm to someone pursuing such cases.

The bench notes these submissions and then asks petitioners to trust the same institutions whose selective inaction had triggered the litigation in the first place. What is lost on the bench is that its procedural answer treats both sides as equals but today, the field of power is as unequal as it can get.

Taken together, the Supreme Court has retreated in its commitment to upholding the secular democratic character of the nation. The court has chosen to absent itself from a field in which its earlier interventions, like in the Pravasi Bhalai Sangathan, the Tehseen Poonawalla, and the Shaheen Abdulla cases had at least kept some institutional pressure on enforcement.

The direction to transmit a copy of the judgment to all High Courts in page 125, with a suggestion that they may consider issuing administrative practice directions or guidelines, is the formal closing of a chapter that ran, fitfully, for over a decade.

The grave contradiction and dangerous precedent

The judgment’s own internal evidence undermines its confidence in the adequacy of existing remedies. Part III, dealing with at least eight contempt petitions of the Court’s earlier directions to the Police for suo-motu registering FIRs when hate speeches are made, records that two matters are closed only because compliance had been achieved, but in at least four others the respondent authorities were granted two more weeks to file responses, with separate listing directed for May 19.

Importantly, two petitions seeking contempt action filed directly without approaching the police with relevant material first were closed as the court stated that inferring automatic contempt would be “overly broad and untenable”.

If Part I is correct that there is no systemic failure of such magnitude as to warrant the court’s continuous monitoring, then Part III cannot sit on the same shelf. The existence of both contradicts each other.

Online hate

The treatment of online hate is equally unconvincing. The Telangana appeal in Part II concerned precisely the kind of digital ecosystem in which hate speech is now industrialised: hashtags, viral clips, algorithmic amplification, the conversion of a single inflammatory statement into a thousand repetitions. The court disposed of that appeal in a single paragraph. For a judgment delivered in 2026, after years of evidence about how communal speech is produced, distributed, and consumed online, there is a remarkable intellectual hole in Justice Nath’s judgment. It registers the problem and walks past it.

Germany’s Section 130 criminalises incitement of hatred against national, racial, or religious groups, calls for violent or arbitrary measures, and certain forms of dignity-attacking dissemination, with penalties of up to five years in serious cases. The United Kingdom’s Public Order Act 1986 criminalises threatening, abusive, or insulting material intended or likely to stir up racial hatred, supplemented by religious-hatred provisions and now by the Online Safety Act’s illegal-content duties. Canada’s Criminal Code names protected groups expressly under Section 319 and criminalises advocating genocide under Section 318.

Of course, none of these systems are without criticism. However, other democracies are increasingly recognising that bare reliance on generic public-order offences is insufficient to deal with online hate speech. They have specified the harm more precisely and institutionalised enforcement more directly.

The Indian Supreme Court was free to refuse to legislate from the bench. It was not, however, obliged to claim that the existing patchwork is sufficient. The Karnataka legislature evidently does not think so. Its Hate Speech and Hate Crimes (Prevention) Bill, 2025, attempts a dedicated definition, makes offences cognisable and non-bailable, prescribes graded imprisonment of one to seven years for a first offence and two to ten for repeat offenders, empowers preventive action, and authorises platform takedown directions through a designated officer.

The Bill has its own constitutional vulnerabilities of breadth, vagueness, executive discretion, and the risk of selective prosecution, and the Governor has even reserved it for the President’s consideration, but its very existence rebuts the easy claim that nothing more specific is needed. It also exposes the hard work to engage with the complicated issue that Justices Nath and Mehta were unwilling to undertake.

Elections and the speeches

What makes the April judgment more than a missed opportunity is the political environment into which it is delivered. The 2024 general election campaign saw the Prime Minister describe Muslims as “infiltrators” who have “more children” and accuse the opposition of planning to redistribute Hindu wealth to them. The Election Commission of India, after much delay, asked star campaigners to refrain from campaigning along religious lines.

India Hate Lab’s 2024 dataset recorded 1,165 in-person hate-speech events for the year, an increase of more than 74 per cent over 2023.

The 2026 State elections deepened the Hindu–Muslim split in voting patterns. In Assam, Himanta Biswa Sarma said that “4-5 lakh Miya votes” (a term used for Bengali-speaking Muslims in Assam) would be removed during the Special Intensive Revision, that his job is to “give trouble to Miya people”, and that the BJP is “directly against Miyas”.

Later, the Supreme Court Chief Justice’s bench declined to entertain a petition seeking to restrain Sarma and pushed the matter to the Gauhati High Court, which has since only issued notice. Meanwhile, Sarma continued giving hate speeches.

This is the background against which Justice Nath’s judgment’s words “no cognisable offence was made out” land. They land in a political economy in which polarising speech is not an unfortunate by-product but a well-thought out organising strategy. They land where dogwhistling has been refined into a method. They land at a moment when the judiciary’s capacity to set norms, through suo motu cognisance, continuing mandamus, even sharply worded oral observations, is one of the few external pressures still available against majoritarian campaign rhetoric.

The Supreme Court has now substantially reduced that capacity and offered the doctrinal language of “judicial restraint” as the justification.

The Constitution’s promise of fraternity is one of its hardest promises. It is also the most easily orphaned, because no single agency owns it.

The hope that the judiciary might at least insist on its enforcement against the most powerful speakers, the Ministers, the Chief Ministers, was always a thin hope. After April 29, it is thinner. Restraint of this kind has a name. Permission.

Saurav Das is an investigative journalist writing on law, judiciary, crime, and policy.

Also Read | A ride with Hate

Also Read | The ‘I Love Muhammad’ controversy in Uttar Pradesh