Vishaka at 28: Guidelines as Legislation, and What POSH Did to Them

There is a particular kind of judicial work the Indian Supreme Court does that no other apex court in a common-law democracy quite matches. It writes law into a void. Not by stretching a statute past its natural meaning, which is ordinary interpretation, and not by reading new rights into a constitutional silence, which most apex courts do at least sometimes. The Indian Supreme Court occasionally writes the operative provisions of a statute itself, declares those provisions binding under Article 141, and tells Parliament to come back when it is ready. Vishaka v. State of Rajasthan is the cleanest example we have. The 1997 guidelines functioned as a working sexual harassment statute for sixteen years. When Parliament finally enacted the POSH Act in 2013, it did not simply codify what the Court had written. It rewrote it, added a defendant-protective spine the guidelines did not have, and slowed the redress machinery the guidelines had set running. The codification was a gain in legitimacy. It was also a loss in protection. We do not say this clearly enough.

What Vishaka actually did

The factual record of Vishaka is now familiar in outline and worth restating in detail because the detail matters. Bhanwari Devi, a saathin in the Rajasthan government's Women's Development Programme, had attempted to stop a child marriage in her village. She was gang-raped in retaliation by upper-caste men of the village in 1992. The trial court acquitted the accused. The High Court's handling of the appeal was widely criticised. A coalition of women's rights organisations, led by an NGO called Vishaka, filed a writ petition under Article 32 — not for Bhanwari Devi individually, because the criminal case lay elsewhere, but for a class declaration: that sexual harassment at the workplace violated the right to equality under Articles 14 and 15, the right to practice a profession under Article 19(1)(g), and the right to life and dignity under Article 21. The petition asked the Court to lay down enforceable guidelines for the prevention and redressal of workplace harassment, given that no statute then addressed the question.¹

What the three-judge bench, speaking through Verma CJ, did next has not, in my reading, been adequately distinguished from the run of public interest jurisprudence around it. The Court did not say sexual harassment violated fundamental rights and direct Parliament to legislate. It did not read sexual harassment into an existing statute by purposive construction. It wrote twelve operative clauses defining sexual harassment, requiring every employer to set up a Complaints Committee headed by a woman, prescribing that not less than half of the Committee should be women, requiring third-party participation from an NGO familiar with the issue, mandating preventive and awareness measures, fixing a duty on the employer to assist in any criminal complaint, and providing for action against the perpetrator. The Court grounded each clause in either a constitutional provision or an article of the Convention on the Elimination of All Forms of Discrimination Against Women, which India had ratified in 1993. It then declared that the guidelines would be binding 'until suitable legislation is enacted' under Article 141.⁴

This was not a default position. It was a deliberate choice between three available options. The Court could have stopped at a declaration of unconstitutionality and asked Parliament to act. It could have read the existing penal provisions — section 354 of the Indian Penal Code, principally — purposively to cover workplace harassment, which would have left employer-side remedy entirely unaddressed. Or it could write a small statute and call it guidelines. It chose the third path because the first would have produced nothing, the second would have produced individual criminal trials but no workplace machinery, and the third produced a working regime overnight.

The architecture: how PIL made this possible

The fact that the Court could do this at all was the cumulative product of a particular institutional turn. Shyam Divan has traced this turn in detail — the post-Emergency Supreme Court's reinvention of itself as an institution of governance, the broadening of standing to allow concerned citizens and NGOs to file on behalf of those who could not, the procedural flexibility of the PIL forum, and the willingness to issue continuing mandamus that converts a judgment into an administrative project the Court itself monitors.¹¹ By 1997, the architecture for what Vishaka would do already existed. The standing question — could an NGO file for a class of working women none of whom were directly before the Court — was effectively settled by the Judges' Transfer case and what followed it. The remedial question — could the Court write detailed prescriptive directions rather than declare rights and remit — had been settled by, among others, Bandhua Mukti Morcha and the bonded labour line of cases.

What was new in Vishaka was that the Court used this PIL architecture to perform a legislative function in its sharpest form. There was no existing statute to interpret, no executive direction to test, no concrete dispute over rights between identified parties. There was only a constitutional vacuum and a class of persons whose rights kept getting violated inside that vacuum. The Court filled the vacuum with operative provisions. I think this is defensible. The Court had to do it because no other organ would, and the constitutional rights in question were not abstract. But I also think the defence has not been honest about what the Court was doing. We have collectively called Vishaka 'guidelines' for nearly three decades, as though that softens the legislative character of the act. It does not. Vishaka was Parliament-work performed by judges under the cover of remedial directions. The reason we should be honest about that is that we cannot otherwise evaluate what the 2013 Act did to it.

The first leg: what the guidelines protected that POSH does not

The 2013 Act, which received presidential assent on 22 April 2013 and came into force in December that year, was politically necessary.³ Long-running judicial legislation is a poor substitute for parliamentary law. The Verma Committee report following the 2012 Delhi gang-rape gave the legislation its final push. None of this is in dispute. What is in dispute, or should be, is whether the resulting statute preserved the protective spine of what it replaced.

Read the two texts side by side and three differences leap out. The guidelines' definition of sexual harassment was tied to two limbs: (i) unwelcome sexually determined behaviour, and (ii) the creation of a hostile work environment or detriment in employment. The Court drafted the second limb because it understood that workplace harassment is rarely a single dramatic event; it is a course of conduct that makes the workplace impossible for the woman to remain in. The POSH Act retains both limbs in section 2(n) and 3, but the implementing machinery — the Internal Complaints Committee structure, the limitation period, the inquiry procedure — is pitched almost entirely at discrete incidents that can be specifically pleaded. A 'hostile environment' over six months does not fit easily into a complaint that must be filed within three months of the incident, extendable to six, with the burden on the woman to specify dates, times, and acts.⁷ The guidelines had no such limitation. They could not, because they were guidelines.

Second, the guidelines made no provision for action against a 'false' complaint. The Court understood, I think correctly, that any such provision would be used not against the rare malicious complainant but against the woman whose complaint the Committee found unproven — which is a very different thing. Section 14 of the POSH Act fills exactly that gap. It authorises the Internal Committee to recommend action against a complainant who makes a 'false or malicious complaint' or produces 'false or misleading' evidence.⁸ The defenders of section 14 will say its first proviso requires more than a mere finding that the complaint is unproven — there must be a finding of malice. They are right about the text. They are wrong about the effect. In a regime where the complaint is heard by an Internal Committee staffed by the employer's own representatives, the deterrent operates regardless of whether the malice finding is ever made. The chilling effect is the provision.

Third — and this is the difference that gets least attention — the guidelines made the employer's duty unconditional. Every workplace was required to constitute the Complaints Committee. The POSH Act conditions this on workplaces with ten or more employees.⁶ Below that threshold the woman is sent to the Local Committee, a district-level body that, even ten years after enactment, exists patchily and processes complaints slowly. The shift looks administrative — it is anything but. The women most vulnerable to harassment in India work in workplaces with fewer than ten employees: domestic workers, small-shop workers, beauty parlours, home-based subcontracting. The guidelines protected them. POSH, by the architecture of section 4 read with section 6, sends them through a parallel and weaker forum.

The second leg: the Court already showed us what codification would do

The story does not begin in 2013. It begins in 1999, two years after Vishaka, when the Supreme Court decided Apparel Export Promotion Council v. A.K. Chopra. The respondent had been dismissed for trying to molest a junior colleague in the office. The departmental inquiry found the charge proved. The High Court reinstated him on the ground that the conduct did not amount to sexual harassment because no actual molestation was completed. The Supreme Court reversed and restored the dismissal. The result was correct. The reasoning was not.

The Court held, in a much-quoted paragraph, that 'any action or gesture which, whether directly or by implication, aims at or has the tendency to outrage the modesty of a female employee must fall under the general concept of the definition of sexual harassment'.² Ratna Kapur has pointed out what is wrong with this — and the point is more serious than it has been credited. The Vishaka guidelines defined sexual harassment as unwelcome sexual conduct creating a hostile work environment or affecting employment outcomes. That definition is rights-based: it protects the woman's right to a workplace she can function in. The Chopra court replaced 'unwelcome sexual conduct' with 'outraging the modesty of a female employee' — a phrase carried over from the colonial-era penal code, premised on the woman as a passive bearer of honour rather than as a rights-holder.¹⁰

The shift is small in words and total in framing. The Vishaka woman is harmed because her right to work in a non-hostile environment is violated. The Chopra woman is harmed because her modesty is outraged. The first analysis can extend to a bar dancer, a waitress, a sex worker — the rights are theirs equally. The second cannot, because modesty as a juridical concept attaches to some women more readily than others. Within two years of Vishaka, the Court's own follow-on jurisprudence had begun rewriting what the guidelines protected. The 2013 Act inherits this drift. Section 3, on the prohibition of sexual harassment, hews closer to the rights-based Vishaka formulation. Section 2(n), the definition, is closer in spirit to Chopra's catalogue of physical advances, requests for sexual favours, and sexually-coloured remarks. The Act keeps both because Parliament could not pick.

The protectionist register that runs through Chopra is not incidental. It is the same register that runs through the section 498A jurisprudence on cruelty by husband or relatives, where the woman is constructed as a passive bearer of harm against whom an outrage is committed, rather than as a rights-holder whose autonomy in a specific institutional setting — the marriage, the workplace — has been violated. Both bodies of law, the post-Vishaka harassment jurisprudence and the section 498A cruelty jurisprudence, started from a rights-based premise and slid into a protectionist one. The slide is not unique to one bench or one fact pattern. It is a feature of how Indian courts metabolise gender-rights claims. The reason this matters for Vishaka at twenty-eight is that the codification we got was drafted against the backdrop of Chopra's protectionist gloss, not against the backdrop of the guidelines' original rights-based architecture. Parliament inherited the gloss as much as it inherited the guidelines, and the gloss is what is most visible in section 2(n).

The counter-argument, addressed honestly

The strongest defence of POSH against the case I have made does not deny these textual differences. It denies that they are losses. It runs roughly as follows. Guidelines without a statutory base are unenforceable in any serious sense — the Supreme Court itself, in Medha Kotwal Lele, spent over a decade monitoring compliance and found that compliance was patchy.⁵ A statute, even an imperfect one, provides a stable platform that does not require a writ petition for every enforcement gap. The limitation period in section 9 is short, but predictability serves both sides — without it, employers face an open-ended risk that itself chills hiring of women into supervisory positions. Section 14 against false complaints is a legitimate procedural safeguard whose actual application is rare. The ten-employee threshold reflects the practical limit of an internal complaints mechanism — beneath it, the woman is not without remedy, she has the Local Committee. The guidelines were never an end-state; they were a placeholder. POSH is the real thing.

This defence has real force. The guidelines did not produce universal workplace compliance. The 2023 decision in Aureliano Fernandes v. State of Goa is precisely a catalogue of how far short of POSH compliance even POSH-bound institutions have fallen.⁹ If guidelines were sufficient, we should have seen the curve bend in the sixteen years between 1997 and 2013. We did not. A statute, with its administrative apparatus, training requirements, annual reporting obligations, district officer structure, and rule-making power, can in principle do what guidelines could not.

I take this defence seriously, and the reason I do not finally accept it is that it conflates two questions. The first is whether codification was needed. The answer is yes — judicial legislation is structurally unstable. The second is whether the particular codification we got preserved the protective architecture of what it replaced. The answer there, I argue, is no. The defence above slides from the first answer to the second as though one entailed the other. It does not. Parliament could have written a POSH Act that retained the unconditional employer duty, dropped section 14 or pitched it at a substantially higher threshold than mere malice, extended the limitation in line with the continuing-offence character of workplace harassment, and pitched the definition exclusively at the Vishaka two-limb test. It chose not to. The choice was political. Calling it codification masks the choice.

There is one further point worth making about the form of judicial work that Vishaka exemplifies, and it bears directly on why the codified text is weaker. The guidelines were possible because the writ jurisdiction under Article 32 permits the Court to fashion remedies appropriate to the right it is enforcing. The fashioning power is broad. A statute, drafted by Parliament, is a different animal. It must be passed by both Houses, survive constitutional scrutiny, anticipate every administrative actor down to the panchayat level, and accommodate every interest group with a seat at the drafting table. The Vishaka bench had to satisfy three judges. The POSH drafters had to satisfy the Ministry of Women and Child Development, the Ministry of Labour, the Ministry of Law, the corporate sector's standing committee submissions, the trade union submissions, the Standing Committee on Human Resource Development, and a Cabinet review. Each of those filters has its own preference. The aggregate of those preferences is what we got.

This is, in one sense, exactly what democratic legislation is for. The Vishaka bench had no electoral mandate to write workplace law; POSH does. But the comparison cuts both ways. The bench had no electoral mandate, and so it wrote a regime that simply asked what the right required and answered. The legislative process has every mandate, and so the law it produced reflects every mandate's preference, including the preference of the regulated employers themselves. The result is that the codified version of Vishaka is, in protective terms, weaker than the un-codified one. This is not a paradox. It is what happens when interest-balancing legislation replaces rights-fashioning adjudication. There is no general lesson here about which mode is better. There is a specific observation: in this case, the courtroom produced a stronger protective text than the legislature did, and we ought to be willing to say so.

One way to test this proposition is to ask what would have happened had the guidelines not been written at all. The counterfactual is not difficult to construct. Bhanwari Devi's criminal case would have proceeded through the appeals as it did. The class question — what workplace harassment is, and what an employer's duty in respect of it is — would have continued to be answered nowhere, because no court has occasion to answer it in an ordinary tort action and no labour statute reached the matter. The Sexual Harassment Bill was first introduced in 2007, ten years after Vishaka, and even then it sat in Standing Committee referral and inter-ministerial consultation for another six. Without the guidelines acting as the floor, those six years would have been six years without a floor. The eight years before that, the same. The first complaints committees in Indian workplaces — and the only complaints committees in many of them through the 2000s — existed because the Supreme Court had said they must. The administrative reality is that the guidelines did the work the statute eventually claimed credit for. To say codification was a gain in legitimacy is true. To say codification was a gain in protection is, on this record, false.

The implementation gap, and the second mandamus

I should say something about Aureliano Fernandes, because critics of the case I am making will point to it as evidence that the problem is implementation rather than text. The 2023 decision is a careful documentation of what has gone wrong with POSH compliance in the decade since its enactment. The Internal Complaints Committees that section 4 requires exist on paper in many institutions and not in fact. The training the rules require is performative where it happens at all. The Local Committees that section 6 establishes for unorganised-sector complaints are under-funded, under-staffed, and in many districts effectively absent. The Court issued fresh directions: a monitoring framework, public reporting of compliance, the strengthening of Local Committees, awareness campaigns. It is, in other words, the same continuing-mandamus instinct the Court applied to Vishaka itself in Medha Kotwal Lele, now turned on Vishaka's statutory successor.

Two readings of this are available. The first — the implementation reading — says the textual differences between Vishaka and POSH are minor; what matters is whether either is actually applied, and neither has been, because Indian institutional culture resists workplace gender accountability. The second — the architectural reading — says the textual differences feed into the implementation gap. A statute that pitches its definitional anchor in 'modesty' rather than 'hostile environment', that conditions employer duty on a ten-employee threshold, that bakes in a 'false complaint' deterrent and a short limitation period, will be under-implemented in different and more harmful ways than guidelines that did none of these things. The implementation reading and the architectural reading are not rivals. They are stacked. POSH has both an implementation problem and a design problem, and the design problem makes the implementation problem worse.

The fact that the Court has had to return — twice, if we count Medha Kotwal Lele as the first return — suggests that the original act of judicial legislation has not, in fact, ended. The guidelines were declared binding 'until suitable legislation is enacted'. Suitable legislation has been enacted. And yet the Court remains in the field, monitoring the legislation, prescribing reporting obligations, ordering audits, and supplementing the statute with directions that look very like additional guidelines. The cycle is unusual but not, on Divan's analysis of continuing mandamus, surprising. Once a court takes responsibility for the administrative implementation of a right, the responsibility does not lapse with the passage of a statute that purports to take over. It transfers from the original guidelines to the statute, but the Court keeps the file open. We have, in effect, a quasi-statutory regime: a parliamentary text that is read in light of, supplemented by, and partially corrected through, ongoing judicial supervision. The result is not as clean as either rule-of-law theory or democratic theory would want. It is what we have.

What survives

Twenty-eight years on, the Vishaka guidelines are formally superseded. POSH governs the field. The Court that wrote the guidelines has, in Aureliano Fernandes, returned to monitor the statute that replaced them — the same continuing-mandamus instinct that produced the guidelines now keeps watch over their codified successor. What survives Vishaka, in my reading, is two things, and they do not run together.

The first is the doctrinal point: the Indian Supreme Court can, in a constitutional vacuum and where rights are otherwise unenforceable, write the operative provisions of what amounts to a small statute and bind the country to them under Article 141. This is a feature of the system. It is also a feature that produces unstable law, and the instability is a real cost. Vishaka was the cleanest case for the exercise of this power because the right was clear, the vacuum was complete, and the legislature had had years to act and had not. It does not follow that the same power should be used wherever the legislature is slow. Most of the cases in which it has been used since are not as clean as Vishaka, and the cost is now accumulating.

The second is the substantive point: when a right-fashioning judgment is later codified by a legislature, the codification is not necessarily a step forward in protection. It can be a step sideways or backwards. POSH is not POSH-minus-the-objectionable-clauses-of-Vishaka. POSH is Vishaka filtered through the institutional logic of a Parliament that has to please the employers as well as the employees. The losses we have absorbed are: the unconditional employer duty, the absence of a complaint-discrediting provision, an indefinite filing horizon, and a rights-based rather than a modesty-based definitional anchor. The gains we have absorbed are: democratic legitimacy, an administrative reporting machinery, statutory rules, and a district officer structure. The gains are not nothing. The losses are not nothing either. The honest position is to count them both.

The accounting matters because the same pattern is likely to recur. There are at least three other areas of Indian constitutional law where the Court has effectively written a small regime in advance of legislation — bonded labour, disability access, environmental clearance — and where, in each case, a statute has eventually followed the directions. In none of those cases has the academic discussion been honest about whether the statute that followed preserved or weakened the directions it replaced. The pattern of dilution is not a Vishaka-specific accident. It is what happens when judge-written regimes get handed to a legislature that has its own constituencies to please. If we are going to defend the Court's continuing power to write in the gap — and I think we should, for the reasons given above — we ought also to keep an accurate ledger of what happens to those directions once Parliament gets around to them. The ledger so far is not flattering to the codifying institution.

The Court that wrote Vishaka was performing a function the Constitution does not, on its face, give it. The Parliament that wrote POSH was performing a function the Constitution does, on its face, give it. The first produced the stronger text. We should sit with that, rather than fold it into the standard story of the Court overreaching and the legislature correcting. In this case, on this question, the Court legislated and the legislature legislated, and what the legislature did was not a correction. It was a softening. Twenty-eight years on, the question worth asking is not whether the Court should have written the guidelines. It is what we ought to have done with them when Parliament finally got around to its own work.

Get the week's biggest decisions, distilled.

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