Can the Supreme Court “ban” a person?

March 11, 2026March 13, 2026 Gautam Bhatia2 Comments

In medieval England, there existed something called a “bill of attainder.” A bill of attainder refers to legislation “that declares an individual or group guilty of misconduct or a crime and imposes punishment without the benefit of a judicial proceeding.” Bills of attainder were enacted by the medieval English parliament to “deal with” people that the legislature really did not like, and their purpose was to visit severe consequences (such as deprivation of property, or even of life and liberty) without the inconvenience of a judicial trial (see here).

Bills of attainder were considered an anathema, and were abolished with the advent of the 18th century. Legislation, by its very nature, was meant to be general in nature, and not targeting specific individuals with punishment (in modern democracies, such a thing could only be done after a trial, conducted under the aegis of an independent judiciary). Modern India saw one curious attempt at a (reverse) bill of attainder, when Indira Gandhi’s voided election to Parliament was sought to be validated by the 39th Amendment. This was struck down by the Supreme Court in Indira Nehru Gandhi vs Raj Narain, on grounds of violation of the basic structure.

Today’s Supreme Court order in the suo motu NCERT textbook case has created a new and paradoxical concept: a judicial bill of attainder. Previously, on this blog, we had discussed the Supreme Court’s order banning the NCERT Class VIII textbook that had a chapter on “judicial corruption,” under the heading “Can the Supreme Court ‘ban’ a book?” Today, the Court has gone beyond banning the book, and has also banned people: in particular, the three individuals who had authored the chapter in question. The relevant part of the Court’s, in paragraph 8, states that:

“At the outset we have no reason to doubt that professor Michel Danino along with Ms Diwakar and Mr Alok Prasanna Kumar either does not reasonable knowledge about Indian judiciary or they deliberately knowingly misrepresented the facts in order to project a negative image of Indian judiciary before students of class 8 who are at an impressionable age. There is no reason as to why such persons be associated in any manner with preparation of curriculum or finalisation of text book for the next generation. We direct union, all states, all institutions recieving state funds, to disassociate them from rendering any service which would mean payment to them from public funds. This is subject to them approaching this court for seeking modification after tendering their response.”

It should be immediately obvious that banning an individual from a certain type of employment is a coercive act that infringes their constitutional rights (in particular, their freedom of trade under Article 19(1)(g) and their right to livelihood under Article 21). The Supreme Court’s order thus raises two important questions of jurisdiction. The first is the exact question that we had raised in our previous post: namely, that the Constitution does not authorise the restriction of fundamental rights through judicial decree. Fundamental rights can be restricted only by law, and judicial orders do not constitute “law” for the purposes of the fundamental rights chapter. The constitutionally-assigned task of the Court is to judicially review legislative action restricting fundamental rights for compliance with the Constitution, not to restrict rights itself (from which, of course, there is no review). In a comment to that post, a reader had pointed out that the book ban might be justified under the Court’s contempt powers; whatever the merits of that view, it is clear that contempt powers do not authorise the Court to deprive an individual of employment.

The second point is that this punishment has been visited upon the three individuals without any clarity about what law they have broken, without a charge or a trial, without a hearing, and without a judgment. In essence, the Supreme Court has created a punishment without a law or a process. In other words, this is a judicial bill of attainder: the punishment of an individual, or individuals, purely through judicial decree. And here is both the paradox and the tragic irony: bills of attainder were deemed to be an anathema precisely because it was widely believed that in a modern society, an individual can be punished only if they are found in breach of a pre-existing general law, by an independent judiciary, after due process. What the Court has done here is to take the form of the bill of attainder, and clothe it with judicial power.

The Court’s order invites the three individuals to apply to the Court for a modification of the order – presumably, to “explain” themselves. This does not make the order legal or palatable. To start with, it inverts the fundamental principle that an individual ought not to be punished before a hearing by first delivering the punishment, and then asking the individual to come and explain why they should not be punished. Even in departmental enquiries, there is first a show-cause notice, then a defence, and finally a verdict. If, tomorrow, a department was to fire an employee and then issue a notice asking the employee to justify why they should not be fired, presumably this very court would strike it down for violating administrative law. And secondly, even a hearing does not cure the fundamental absence of jurisdiction, and the fact that this remains, at its core, a judicial bill of attainder: punishment by judicial decree, without law or legal process.

Banning books and banning people is not what constitutional courts are established for, and they form no part of a constitutional court’s remit – no matter how offensive the court may find a particular book, or how odious it may find a particular person. Judicial outrage cannot be a substitute for constitutionalism; to make it so would be to set both the rule of law, and constitutionalism, adrift without anchor.

The legality and constitutionality of the SIR

March 10, 2026 Vasudev DevadasanLeave a comment

The previous post (here) discussed the Supreme Court’s flawed approach when hearing the legal challenge to the Special Intensive Revision (‘SIR’). This post discusses the substance of the legal challenge to the SIR. The post first outlines the statutory scheme governing the revision of rolls. The post then examines the current SIR exercise against this scheme, which raises several questions regarding the legality and constitutionality of the SIR. The post discusses each of these questions in turn.

Before beginning, a clarification on scope is warranted. The SIR raises several systemic legal problems including indirect discrimination and widespread disenfranchisement. The exercise has also led to numerous operational and political issues, such as the incredibly short timelines imposed by the ECI, the organised use of Form 7s to delete Muslim voters, and the use of political power to influence adjudicatory processes under the SIR. The SIR also revives questions regarding the independence of the Election Commission of India (discussed on this blog here). This post does not deal these issues. Instead, it focuses on the doctrinal arguments for invalidating the SIR notifications in the case reserved for judgment by the Supreme Court.

Statutory Scheme for Revision of Rolls

Under s 21 of the Representation of the People Act 1950 (‘RPA’), the Election Commission of India (‘ECI’) is tasked with maintaining updated electoral rolls in “each constituency” in the country. Section 21(2)(a) states that rolls shall be revised “in the prescribed manner” before each general election to the Lok Sabha or State Legislative Assembly and before each bye-election. The use of the phrase “in the prescribed manner” makes it clear that the rolls shall be revised in accordance with statutory rules under the RPA, namely the Registration of Electors Rules 1960 (‘Registration Rules’).

The key provisions concerning the SIR are ss 21(2)(b) and 21(3), and they are worth setting out in full. Section 21(2)(b) states that the electoral roll:

“shall be revised in any year in the prescribed manner by reference to the qualifying date if such revision has been directed by the Election Commission”

While s 21(3) states:

“Notwithstanding anything contained in sub-section (2), the Election Commission may at any time, for reasons to be recorded, direct a special revision of the electoral roll for any constituency or any part of a constituency in such manner as it may think fit:”

Both ss 21(2)(b) and 21(3) empower the ECI to revise the rolls at any time. However, s 21(2) requires the rolls to be revised according to the Registration Rules (“in the prescribed manner”) while s 21(3) omits this requirement. However, the power in s 21(3) is qualified in three distinct ways. First, the process under s 21(3) is referred to as a “special” revision, indicating the power is an exceptional one. Second, while s 21(2), by reference to s 21(1), refers to revision of rolls in “each constituency”, s 21(3) refers to revisions in “any constituency or any part of a constituency,” implying a geographic limitation, i.e., the power is not for every constituency. Third, s 21(3) requires the ECI to justify its reasons for invoking the provision in writing.

Also note that s 21(3) refers to a “special revision”, but the current exercise is titled a “special intensive revision.” The distinction between a regular revision of rolls and an intensive revision of rolls is found in the Registration Rules. Rule 25(1) states that a revision of rolls “under sub-section (2) of section 21” shall either be a summary revision or an intensive revision, or a part summary part intensive revision. Rule 25(2) states that an intensive revision is the preparation of rolls “afresh” (i.e., discarding existing rolls). The Rule also states that when rolls are revised intensively, Rules 4-23 of the Registration Rules apply. Rules 4-23 set out the various procedures and safeguards the ECI and its agents must abide by when revising rolls.

The Present SIR Exercise

It is important to recognise that the SIR notifications (first for Bihar and then other states) lay down their own process for the revision of rolls. The notifications make references to the RPA and Registration Rules, but also make crucial deviations from them. For example, r 8 of the Registration Rules empowers Election Registration Officers (‘EROs’) to send out and require people to fill out Form 4s. These Forms require only three details: (i) name; (ii) name of father/mother/spouse; and (iii) age. Further, Form 4s require the filler to self-declare their citizenship, with the onus on EROs to subsequently verify suspect cases.

This stands in stark contrast to the SIR notifications under which all electors are required to fill out enumeration forms, and failure to submit these forms results in exclusion from the roll. The notification states, “in case enumeration form is not submitted before [date], the name of the elector can not be included in the draft rolls.” Further, all voters registered after the last intensive revision in the State (2003 in Bihar) must submit additional documentation demonstrating citizenship. Nowhere do the Registration Rules, whether the revision is regular or intensive, contemplate citizens’ exclusion from the rolls for failure to submit Form 4s or an enumeration form, and past ECI guidelines placed the onus on Commission staff to go door-to-door to verify voters. Thus, the SIR notifications contemplate a materially different, and considerably more exclusionary, process than the Registration Rules.

Questions raised by the SIR

The ECI acknowledges that the SIR process is different from the regular revision of rolls ordinarily carried out by the Commission. However, it justifies this novel procedure on the grounds that: (i) art 324 of the Constitution gives it inherent and free-standing powers to ensure the integrity of the voting rolls; and (ii) as the SIR is conducted under s 21(3) of the RPA, the ECI is free to devise its own procedures and the Registration Rules do not strictly apply.

This raises the two questions. First, can the ECI directly rely on art 324 of the Constitution as a source of power for the SIR exercise? Second, can the ECI rely on s 21(3) of the RPA at any point? Or must revision of rolls ordinarily be under s 21(2) and s 21(3) only invoked in exceptional circumstances? And is the pan-India SIR such an exceptional circumstance? Beyond this, there is a third question that must be answered, which is, even if the ECI can resort to s 21(3) to create a new procedure for revising rolls, does this new procedure comply with Constitutional Fundamental Rights and administrative law standards applicable to discretionary State action? These questions are at the heart of the legality and constitutionality of the SIR.

The ECI’s powers under art 324

The ECI’s arguments regarding art 324 can be taken at two levels. The maximalist argument would be that the SIR exercise is carried out entirely under art 324 of the Constitution, which tasks the ECI with “superintendence, direction and control of the preparation of the electoral rolls…”, and thus statutory text and rules are irrelevant. However, this legal position was squarely rejected by the Constitution Bench in Mohinder Singh Gill v. Chief Election Commissioner (1977).

Justice Krishna Iyer observed that while art 324 does empower the ECI to supervise the conduct of elections, art 327 read with entry 72 (List I) of the Seventh Schedule empowers Parliament to enact electoral laws while entry 37 (List II) empowers state legislatures to enact electoral laws. This constitutional scheme led the Court to conclude that if Parliament or a state legislature has passed a law, the ECI must comply with it, and the Commission can only rely on art 324 in “areas left unoccupied by legislation.” This position has been consistently reiterated by the Supreme Court, in cases such as Kanhiyalal Omar v. R K Trivedi (1985), Digvijay Mote v. Union (1993), Kuldip Nayar v. Union (2006), and PIF v. Union (2018). Thus, the ECI cannot rely on art 324 to override existing statutory procedures governing the revision of rolls.

A subtler reading of the ECI’s position would be that the ECI is embarking on a novel exercise (i.e. the pan-India SIR) and because existing statutory procedures do not cover all aspects necessary to undertake such an exercise, the ECI can resort to art 324 to fill in the gaps within existing statutory procedures. The problem with this argument is that, as shown above, the SIR Notifications do not fill in the gaps within existing statutory procedure, but rather create a novel procedure of their own. The RPA and Registration Rules expressly contemplate an intensive revision for “each constituency” that the SIR departs from. Further, in AC Jose v. Sivan Pillai (1984), the Supreme Court expressly found that the ECI cannot authorise new conduct through notifications where the field was already covered by statute, and subsequently claim that the pre-existing statutory procedure did not cover the new conduct, thus warranting resort to art 324’s residuary powers.

In AC Jose, the ECI issued a notification authorising the use of electronic voting machines. At the time, neither the Representation of the People Act 1951 nor the Conduct of Election Rules 1961 authorised the use of voting machines. The Court struck down the ECI’s notification and directed re-polling wherever voting machines had been used. Overruling the High Court judgment that upheld the ECI’s notification, the Supreme Court observed,

“The High Court fell into an obvious fallacy by acceptance of the position that the direction of the Commission was intended to operate in an uncovered field. When the Act and Rules prescribe a particular method of voting, the Commission could not innovate a new method and contend that the use of the mechanical process was not covered by the existing law and, therefore, did not come into conflict with the law in this field.”

Parliament subsequently amended the statute and rules to expressly permit the ECI to conduct elections using electronic voting machines.

The RPA and Registration Rules are an existing statutory framework covering the field of revision of electoral rolls, including intensive revisions. The doctrinal position makes it amply clear that the ECI cannot rely on art 324 to either override this framework or create a novel process and then argue that the existing framework does not cover the new process. If the ECI was truly of the belief that existing procedures could not support a pan-India SIR, Parliament should revise the law. But the ECI cannot supplant the existing statutory scheme for revision of rolls under art 324.

Reliance on s 21(3) of the RPA

In addition to art 324, the ECI also argued that it is justified in diverging from the Registration Rules because s 21(3) of the RPA empowers the Commission to conduct a special revision “as it may think fit” and not in “the prescribed manner.” It is true that s 21(3) clearly grants greater procedural discretion to the ECI in how to carry out the special revision. But as noted above, s 21(3) is an exceptional power, to be applied across a limited area, and justified by reference to written reasons.

Reading the ss 21(2)(b) and 21(3) together makes it clear that these are not two parallel procedures that the ECI is free to choose from at will. Section 21(2)(b) is the ordinary procedure where the Registration Rules unquestionably apply, and s 21(3) is a limited exceptional power to address circumstances that cannot be dealt with under s 21(2). If Parliament wanted to allow the ECI to regulate its own procedures for revision of rolls, it would not have insisted that the revision of rolls under s 21(2) be carried out “in the prescribed manner.” Crucially, if Parliament wanted to allow the ECI unfettered discretion to choose between ss 21(2) and 21(3), it would not have required the ECI to justify its reasons for invoking s 21(3). This makes it clear that there is a legal burden that the ECI must meet before being entitled to resort to the exceptional powers under s 21(3) and create its own procedures for the revision of the rolls. The fact that the ECI has sought to rely on this exceptional power to revise every roll in the country only raises the bar for its justification, as this requires showing why in no part of the country is the ordinary process of roll revision viable.

This brings us to the current SIR. During oral arguments, the ECI repeatedly stressed that because the current exercise was an intensive revision, it warranted novel procedures, including excluding citizens from rolls if they did not fill out enumeration forms. However, the r 25 of the Registration Rules expressly contemplates an intensive revision under s 21(2) of the RPA, dispelling the notion that it is the intensive nature of the revision that warrants resort to the exceptional powers under s 21(3). In fact, the ECI would have been better placed to conduct the entire pan-India SIR under s 21(2)(b) as the provision deals with the preparation of rolls for “each constituency” and the ECI would not have to argue that the words “any constituency” in s 21(3) can include all constituencies. However, if the ECI had conducted the SIR under s 21(2)(b), they would have had to do so “in the prescribed manner” and could not have resorted to the exclusionary enumeration forms.

The ECI’s notification discloses two further potential justifications for the current SIR: (i) rapid urbanisation and frequent migration; and (ii) the need to verify citizenship. The problem here is that Rule 21A of the Registration Rules expressly grants EROs the power to investigate and delete names of persons who are dead, not living in the constituency, or not entitled to be registered (which could plausibly include not being citizens). If s 21(3) ought to be invoked only in the face of circumstances where existing procedures are inadequate, domestic migration and international immigration are not such circumstances. Both phenomena are longstanding and ongoing, and the ECI has carried out innumerable roll revisions in accordance with the Registration Rules despite these phenomena.

The ECI’s reasons for why it needs to deviate from existing statutory procedure and resort to the exceptional power under s 21(3) simply do not pass muster. What we have here is an exercise that, for all intents and purposes, could have been carried out under s 21(2)(b) and the Registration Rules, making the use of s 21(3) unjustified. Put more cynically, the only reason for the ECI to resort to s 21(3) was for it to circumvent complying with the Registration Rules.

Judicial Review Beyond the Statute

The decisions in Mohinder Singh Gill, and more recently in ECI v. Ashok Kumar (2000), clearly establish that the Commission’s powers, whether under art 324 powers or statute, are subject to judicial review on Fundamental Rights and administrative law grounds. Thus, even if the ECI can resort to both art 324 and/or s 21(3) of the RPA, this still leaves the question of whether the ECI’s actions comply with these legal standards.

During oral arguments, the Petitioners argued that the SIR notifications, by presumptively excluding all existing voters unless they submit an enumeration form, violate the precedent in Lal Babu Hussein v ERO (1995). The case concerned an ECI notification requiring EROs to verify the citizenship of individuals based on reports collected by the police. Several of these reports cast doubts on the citizenship of individuals and recommended their removal from the rolls. The Three-Judge Bench of the Supreme Court observed that names were being struck off the rolls based on unquestioned reliance on the police reports and without a notice and hearing being provided to the voters, despite their names being present on prior voting rolls. This led the Three-Judge Bench to hold that where:

“the name is already entered [on the voter roll], it must be presumed that before entering his name the officer concerned must have gone through the procedural requirements under the statute. This would be so even if we invoke Section 114(e) of the Evidence Act. But then possibilities of mistakes cannot be ruled out. These mistakes, if any, would have to be corrected […] after giving the person concerned a reasonable opportunity of being heard.”

The Court further held that the person whose name is sought to be removed on the ground that they are not a citizen must be informed of the basis on which their citizenship has been doubted. The effect of Lalu Babu Hussein is that where a person has already been included on the voter rolls, and their name is sought to be deleted, they must be given notice of why their name is sought to be deleted and a hearing to argue against deletion.

The SIR notification runs afoul of this precedent by providing for the automatic deletion of all individuals who do not submit the enumeration form. It stands to reason that if a person cannot be excluded from the voter roll mechanically, even where they may not be a citizen, they cannot be excluded from the roll mechanically for no reason whatsoever (other than not filling out a form). If the ECI believed that certain voters were non-citizens or no longer residing in the constituency, EROs should have sent notices to only these individuals, rather than presumptively deleting any person who did not submit the enumeration form with appropriate documentation.

In addition to Lal Babu Hussein, the SIR also engages art 14 (the right to equality) in notable ways. The notifications differentiate between citizens who were added to the rolls during the last intensive revision (2003 in Bihar) and voters who were added to the rolls subsequently through ordinary revisions. The latter class are required to submit various government documents with their enumeration forms, while the former class only have to submit enumeration forms. The ECI justified this differentiation by arguing that voters added pursuant to the previous intensive revision enjoy a “presumption of citizenship”, while those added subsequently do not. In West Bengal, the ECI further differentiated between voters, sending voters with “logical discrepancies” notices for additional verification. The discrepancies included situations such as where more than six voters added after the last intensive revision had a common parent.

Now the right to equality requires (i) that the State make a classification using intelligible differentia; and (ii) the classification has a reasonable nexus with the object of the State’s action. As noted above, the SIR makes various distinctions between voters, arguably on intelligible differentia. The ECI’s stated objective for the SIR is to ensure the integrity of the electoral rolls in the face of domestic migration by citizens and international immigration by non-citizens. This leaves the question of whether the distinctions made by the ECI have a nexus with the Commission’s stated aims.

On citizenship: As noted in Lal Babu Hussein, the fact that a person is already on the electoral roll creates a presumption that they were added correctly and are citizens, a presumption that can only be displaced by an investigation and a hearing. Therefore, the ECI cannot suggest that the citizenship of those added to the rolls after 2003 is somehow less secure than those added prior to 2003, especially as it was the ECI itself that added these voters to the rolls after 2003. Therefore, it cannot be said that the ECI’s differentiation of pre and post 2003 voters has any reasonable nexus with the goal of eliminating non-citizens from the rolls as all voters on the rolls enjoy the same presumption of validity. In other words, there is nothing about post-2003 voters that makes their citizenship more legally suspect. Further, the list of documents the ECI requires post-2003 voters to furnish are themselves not definitive proof of citizenship. They include documents such as: (i) identify card of a government office; (ii) matriculation certificates; (iii) family register; and (iv) land allotment certificates by governments. Thus, even the ECI’s heightened scrutiny of post-2003 voters cannot be justified vis-à-vis citizenship tests. In this light, the SIR looks more like a restrictive voter-ID law to disenfranchise marginalised groups than a bona-fide exercise to ensure the integrity of the voter rolls.

On migration and discrepancies: The ECI’s justification for differentiating between pre and post 2003 voters is even less reasonable with respect to its stated aims of addressing voter migration. The ECI has led no evidence that voters registered prior to 2003 are less likely to migrate out of the constituency than newer voters. In fact, one would assume that the most recently registered voters are the least likely to have migrated out of the constituency. Yet, the ECI has placed greater burdens on these newer voters.

The above analysis applies equally to the distinction between voters falling within the ECI’s logical discrepancy lists and those falling outside these lists. There is absolutely nothing to suggest that voters who are one amongst several siblings are either less likely to be citizens or more likely to have migrated and thus warrant greater scrutiny. Thus, even if the Supreme Court were to find that the ECI has carte blanche powers under art 324 and s 21(3) of the RPA, the SIR notification would still violate existing Supreme Court precedent and the right to equality.

Conclusion

As noted at the start of this piece, many of the problems flowing from the SIR are not legal. As the previous post observed, the Supreme Court has itself (erroneously) spent much time addressing numerous operational and political problems stemming from the SIR. This includes passing truly sweeping directions, such as requiring West Bengal’s judicial officers to deal with SIR claims, leading to a slowdown in the regular administration of justice across the state. However, as this post demonstrates, the SIR does suffer from significant legal infirmities that warrant its invalidation. In striking down the SIR, the Supreme Court can put an end not just to the legal controversies but also to the administrative, political, religious, and very human problems created by an ill-timed, ill-conceived, and illegal exercise.

The Return of Galileo: Protecting Online Speech and Delegated Proportionality in the Kenyan Court of Appeal’s BAKE judgment

March 9, 2026 Gautam BhatiaLeave a comment

The judgment of the Kenyan Court of Appeal in Bloggers Association (BAKE) of Kenya vs Attorney General, delivered on March 6, represents an important advance in Kenyan free speech jurisprudence. At issue was the constitutional validity of Sections 22 and 23 of the Computer Misuse and Cybercrimes Act, 2018. These sections criminalised the “intentional publication of false, misleading or fictitious data or misinformation with the intent that the data or information shall be considered or acted upon as authentic” and “publish[ing] information that is false in print, broadcast, data or over a computer system that is calculated or results in panic, chaos, or violence.” While the High Court of Kenya had upheld the constitutionality of these provisions, the Court of Appeal struck them down (see the analysis by Joshua Malidzo Nyawa here).

The case thus raised the issue of whether the State can criminalise bare falsehood, without the added requirement of a prohibited impact (say, for example, propaganda for war, or incitement, or public disorder). The Court of Appeal held that it could not, on the basis that “in a world without universal truths or falsities, the offences may be difficult to prove.” (para 141) It went on to note that:

… history teach[es] us that what we may hold to be false today may turn out to be true tomorrow. For instance, Galileo Galilei was tried and convicted by the Roman Inquisition for stating the truth (that the earth revolves around the sun) in an era where the correctness of his statement was deemed a falsity. These provisions risk criminalising satire, opinions and journalistic inaccuracies. (para 142)

It is important to carefully unpack these two observations, as they make two overlapping – yet distinct – claims. The first is the postmodernist-inflected claim is that the notions of “true” and “false” break down under sustained scrutiny, as this is a “world without absolutes.” While I have no stand on this as a philosophical matter, I do not think it is a constitutionally correct position. In certain contexts, the law does require determinations of “truth”: the most obvious example is defamation law. In other words, whatever philosophy has to say on the subject, the law does presume that “true” and “false” are not entirely subjective constructs.

The observation in paragraph 142, however, is both more modest and – in my view – more accurate. That is, whatever the ontological status of “true” and “false,” there are excellent reasons why we do not vest the power to determine truth and falsity in the State, backed up with coercive sanction. The example of Galileo demonstrates the risk of a constitutional order in which the State (or the Church’s) version of what is true becomes the truth, because all other versions are persecuted out of existence. Indeed, in his judgment in Kunal Kamra vs Union of India, while examining a very similar law, Patel J. noted:

… while some things may be absolutely true (that 2+2 = 4), the question before us is whether the State can arrogate to itself the power to determine what, outside the starkly obvious, may be true or false. History has no shortage of examples; the most immediate one may be Galileo. (para 165)

I would therefore read the Court of Appeal judgment as prohibiting the State from using its coercive powers in service of determining one “truth.”

A third point also follows from para 142, which is the risk to criminalising satire or opinions. Ostensibly, it could be argued that a provision dealing only with truth or falsity would not pose any risk to opinion. However, this is not true, as there is a whole range of expressive content where the question of whether it is a fact or an opinion is itself contested. The most striking example of this is the South African Constitutional Court judgment of DA vs ANC, where a bulk election text message stating that the-then President Jacob Zuma “stole” public money caused controversy. While Zuma had never been convicted of theft, the DA argued that the word “stole” was not to be understood as referring to the criminal offence of theft, but as referring to a broader accusation of misuse of public funds for personal gratification. On the narrow meaning of the word “stole”, the text message was clearly a false statement. On the broader meaning of the word “stole”, the text was – equally clearly – an expression of opinion. The difficulty of determining this is highlighted by the fact that the issue resulted in multiple split judgments by the South African High Court, the Supreme Court of Appeal, and the Constitutional Court. What this shows, therefore, is that the Kenyan Court of Appeal was entirely correct in noting that while a law may claim to criminalise only bare falsehoods, and therefore not touch “opinion”, it is impossible to limit the reach of the law to the former without impacting the latter. Consequently, the Court was right to note that the provisions were unconstitutionally “broad, wide, untargeted, akin to unguided missiles, and likely to net innocent citizens.” (para 141) (for similar arguments, see the present author’s written submissions in Kunal Kamra vs Union of India, Part I and Part II).

The BAKE case also raised issues of online privacy, which we may consider briefly. There was a range of provisions in the Act designed to prevent and prosecute cyber-crime, with a knock-on effect on privacy rights. The Court of Appeal upheld these provisions, on two bases: first, that for the most part, the coercive powers granted to the police were for the most part ring-fenced by the requirements of judicial sanction, and specific, non-general warrants; and secondly, that the provisions themselves encoded the proportionality standard. It is difficult to argue with this finding, as the law cannot go much further than this. The Court of Appeal was cognisant of the remaining potential of misuse, and accordingly observed that it was the task of the first-instant judiciary – that had been granted powers under the Act – to check abuse.

One can think of this as a form of delegated proportionality: that is, proportionality is to be deployed not only in constitutional challenges by constitutional courts, but by first-instance courts whenever State action affects individual rights. It is important to note that the Kenyan courts are no strangers to this idea: in 2023, Senior Principal Magistrate M.A. Opondo applied proportionality to the question of police remand. While no label has yet been given to this phenomenon, I think that “delegated proportionality” captures its essential aspects; and we shall see if, going forward, Kenyan magistrates apply it to the privacy-infringing provisions of the Act.

I would further argue, in fact, that the Court of Appeal’s observations provide a path forward should this not happen. If, over time, it is found that Magistrates are failing to apply proportionality in a majority of the cases before them, then it can potentially be argued that a “unconstitutional state of affairs” has come into existence, which engages the supervisory jurisdiction of the constitutional courts. Of course, it is to be hoped that this is never needed.

A brief, final note of disagreement: the Court of Appeal upheld the criminalisation of “grossly offensive” speech by equating it to “online harassment.” However, “offensive” and “harassing” are two very different concepts. While one can appreciate the Court’s attempt to “read down” the provision, the gap between the two remains glaringly large, and raises the spectre of the State actually doing the harassment in question, by going after bloggers or social media posters who are merely being “offensive.” This will be one to watch for the future.

Other than this small quibble, however, the Court of Appeal’s judgment is soundly reasoned, and exceptional for being almost entirely unexceptionable. If more judgments were written like this, us critics would soon have very little to write about!

Can the Supreme Court “ban” a book?

February 26, 2026 Gautam Bhatia3 Comments

Today, the Supreme Court passed an order banning an NCERT Class VIII textbook that had a section on “corruption in the judiciary.” These proceedings have generated a substantial amount of heat and noise – not least because the Supreme Court initiated a suo motu case after certain senior counsel urged the Chief Justice to take action. This blog post does not engage with the content of the textbook, but with a crucial anterior question that needs to be considered, which is whether the Supreme Court has the constitutional power to ban books in the first place.

This question was not raised in the proceedings before the Court, because all parties seemed to agree with each other (and with the Court) on the issue. Besides, as we have seen multiple times on recent occasions, the question of whether or not the Court has the jurisdiction to engage in certain actions has vanished from judicial consideration. The question remains important, however, for two reasons: one, as with all other constitutional authorities, the Court does not possess unlimited power. It is crucial to seriously consider what constraints the Constitution places upon judicial power. And two, asserting the power to ban a book is not limited to the facts of this case. The power of censorship in a democratic society is a power of vast scope and proportions, and it is therefore crucial to examine when, under what circumstances, and by whom, it can be exercised.

In fact, the question of whether the judiciary can directly restrict or censor speech has been considered multiple times on this blog, ever since the Court’s 2016 order mandating playing the national anthem in cinema halls. The answer is very straightforward, and it is: no.

The legal reasoning for this is simple, and can be set out schematically.

Article 19(2) of the Constitution only permits restriction upon speech to be done through a “law” that is made by the “State.”
It has long been established that judicial decisions do not constitute “law” for the purposes of Article 19, and the judiciary does not constitute “State” for the purposes of the bill of rights (Mirajkar’s Case; Rupa Ashok Hurra’s Case).
Consequently, the judiciary does not have the constitutional power to directly restrict speech (that is, to ban books or censor films via judicial decree).

Apart from the plain text of the Constitution, there is a deeper reason, grounded in the separation of powers, for why the Constitution does not vest this power in the judiciary. I made this argument on this blog in 2016, and the passage of a decade has yielded no particular reason why it does not hold. I set out the argument once again:

As the Supreme Court held in Kharak Singh vs State of UP, if the State action is to be upheld against Part III claims, the State must “satisfy that… the fundamental rights are not infringed by showing that there is a law and that it does amount -to a reasonable restriction. within the meaning of Art. 19 (2) of the Constitution.” The phrase “there is a law” is crucial, because it sets up a threshold safeguard for the protection of fundamental rights. Plain executive action cannot infringe fundamental rights, even if it is “reasonable” within the meaning of Article 19(2). This is because “law”, which ultimately traces its authority to Parliament (whether it is in the form of legislation, or delegated law-making, such as rules or regulations), envisions a public, deliberative process during which – presumably – civil liberties concerns are taken into account at the time of framing. And after the State makes the law, the Constitution envisages a second layer of safeguards, in the form of judicial review. After the State makes the law, aggrieved citizens can approach the Courts arguing that it violates their fundamental rights … by engaging in direct judicial censorship, the Court short-circuits this crucial two-step safeguard, and bypasses Parliament altogether. By directly restricting speech, it ensures that the deliberative process envisaged by the Constitution when it requires the State to “make a law” under Article 19(2) is rendered chimerical. This is why such judicial action violates the separation of powers.

It should therefore be clear that judicial orders banning books are without jurisdictional foundation, and incorrect in law. While such analysis may have no purchase in the practice of the Court, that is only a reason to critique the Court’s practice, and to re-assert that the exercise of power must always conform to principle.

Geofence Warrants and the Fourth Amendment After Carpenter: The Stakes in Chatrie v. United States [Guest Post]

February 24, 2026 Gautam BhatiaLeave a comment

[This is a guest post by Rudraksh Lakra.]

On January 16, 2026, the Supreme Court of the United States (“SCOTUS”) agreed to review the constitutionality of geofence warrants in Chatrie v. United States (“Chatrie”). A geofence warrant allows law enforcement to collect data from devices within a specified geographic area, known as a geofence, during a designated time period. The petitioner challenged a decision of the Court of Appeals for the Fourth Circuit, which held that the government’s use of a geofence warrant did not constitute a “search” under the Fourth Amendment, as well as the Fourth Circuit’s subsequent en banc decision reaffirming that ruling.

This blog examines the constitutionality of geofence warrants under the Fourth Amendment in Chatrie. The discussion proceeds in three parts. First, it reflects on the rise of digital surveillance and situates geofence warrants within this broader context, explaining how they operate in practice.. Second, it outlines and analyzes the arguments advanced by the government and the petitioner. Third, it situates these arguments within the SCOTUS decision in Carpenter v. United States (2018) (“Carpenter”), examining whether that case should be read narrowly or as signalling a broader shift in digital privacy jurisprudence.

Background

Since the advent of the internet, the digital traces individuals leave behind have become increasingly detailed and enduring, a trend intensified by smartphones, wearable technologies, and the expansion of Internet of Things ecosystems. In this environment, the accumulation of vast repositories of data has become nearly inevitable. The granularity and scale of such data challenge conventional understandings of privacy and what constitutes a reasonable expectation of privacy. Large datasets enable sophisticated data mining and algorithmic analysis, allowing insights to be drawn that would be difficult, if not impossible, to derive from isolated data points. These concerns are amplified when multiple datasets are combined, revealing patterns and inferences far removed from the purposes for which the data was originally collected.

In this context, one of the key defences of large-scale surveillance programmes has been that, while the initial sweep may be broad, the eventual search is narrow, targeted and backed by safegaurds (see ECtHR, Big Brother Watch v. UK and Centrum för Rättvisa v. Sweden). Under this logic, the mere existence of large databases is treated as unproblematic, with constitutional scrutiny directed primarily at the results of a search rather than the scope of the data sweep itself.

Geofence warrants, sometimes referred to as reverse location warrants, raise even more serious constitutional concerns because they depart from this model. Unlike traditional warrants, they are not tied at the outset to a known individual, device, or specific property. Instead, they authorise the collection of location data for all devices within a defined geographic area during a specified time window. Only after this broad collection does law enforcement sift through the data to identify particular individuals of interest. This inversion of the investigative process, in which suspicion follows rather than precedes data collection, raises profound constitutional concerns.

These Geofence warrants requests are typically directed at major technology companies that store large amounts of user data through services such as mobile operating systems, email platforms, and cloud storage. In the United States, both state and federal law enforcement agencies have increasingly relied on private technology companies such as Google, Apple, and Microsoft for user data in criminal investigations. Google receives the vast majority of geofence warrants because it maintains extensive historical location records through its “location history” feature, which are stored in a database known as Sensorvault. Between 2017 and 2018, Google experienced a 1,500% increase in requests. By 2021, geofence warrants constituted 25% of all warrants submitted to Google.

The SCOTUS’s review in Chatrie marks its first opportunity to directly address geofence warrants, and federal Courts of appeals are currently divided on their constitutionality. Before examining the constitutional issues, it is important to understand how geofence warrants function in practice. Geofence warrants generally unfold in a three-step process. For instance, in Chatrie, the process involved:

Step 1 – Initial Geofence Data: Law enforcement obtained a geofence warrant requiring Google to identify all devices detected within a 300-meter radius of the bank during a one-hour period. To comply with this request, Google had to search through its entire location database. The information provided at this stage was anonymized.
Step 2 – Expanded Location Data: After identifying a subset of potentially relevant nine accounts, Google provided additional anonymized location data covering a longer, two-hour period. At this stage, the geographic limits were lifted, meaning the full location history of those accounts during the two-hour window was disclosed. This data is also anonymized.
Step 3 – Deanonymization: At the third step, after law enforcement reviewed the additional location data and narrowed the list of relevant devices to three, Google provided identifying information for specific accounts, including usernames and email addresses.

Importantly, the warrant is only taken at the first stage, meaning law enforcement does not seek a separate warrant before obtaining identifying information in the second and third stages.

Fourth Amendment Knots

The government contended that the petitioner lacked a reasonable expectation of privacy because he voluntarily shared his cellphone location data with Google by opting into the location history service. By doing so, the government contended, he relinquished any Fourth Amendment protection in that information under the third-party doctrine, which generally holds that individuals have no constitutional privacy interest in data they knowingly disclose to a third party. Framed this way, the case becomes less about surveillance and more about assumption of risk in the marketplace. The government attempted to distinguish this case from Carpenter, a decision on which the petitioner heavily relied. In Carpenter, the Supreme Court held that the third-party doctrine did not apply to the government’s extensive and highly detailed collection of historical cell-site location information(“CSLI”), and that obtaining such data constituted a search under the Fourth Amendment.

According to the government, two key differences set this case apart from Carpenter. First, it involved the collection of seven or more days of location data, which the Court described as providing an “intimate window into a person’s life.” By contrast, the geofence warrant here covered a limited geographic area and a short time frame. As a result, the government argued, the warrant did not generate an “all-encompassing record” of the petitioner’s movements or a comparably revealing portrait of his private life. Rather, it simply placed the petitioner near the credit union around the time of the robbery. Second, the government stressed that, unlike the CSLI in Carpenter, Google’s location history data is gathered only when users take the affirmative step of opting into the service.

The petitioner argued that the geofence warrant was inherently unconstitutional. They posited that Carpenter’s reasoning applied squarely to this case. The petitioner emphasized that location history data obtained through a geofence warrant is even more comprehensive than the CSLI at issue in Carpenter. Unlike CSLI, which typically tracks a single individual, geofence warrants enable surveillance of everyone within a defined geographic area. They further highlighted that, by using a geofence warrant, law enforcement can obtain information about all individuals present at sensitive locations such as abortion clinics, political protests, or party conventions at a particular time. Like CSLI, location history data can reveal deeply intimate details about a person’s life. In fact, it is even more invasive, as Google tracks a user’s day-to-day physical movements using GPS, Wi-Fi, mobile networks, and cell tower data at roughly two-minute intervals. Moreover, the petitioner argued that accessing such data is relatively easy and inexpensive for law enforcement, increasing the risk of overuse and abuse.

Alternatively, the petitioner contended that this specific geofence warrant constituted an unlawful search because the government obtained a warrant only for the first stage of the process, but not for the data collection that occurred during the second and third stages.

The petitioner also relied on the decision of the Court of Appeals for the Fifth Circuit in United States v. Smith (2024), which held that geofence warrants are per se unconstitutional. The Court reasoned that the first step of the geofence process requires Google to search through its vast database, reportedly containing location data for over 590 million users, to identify devices present at a particular place and time, even though law enforcement does not yet know, and may never know, whom they are looking for. The Court likened this sweeping search to the “reviled general warrants and writs of assistance” of the colonial era.

The petitioner further argued that the government’s reliance on the third-party doctrine was misplaced for two reasons. First, although Google formally requires users to opt in to location history, the company’s design strongly nudges users toward enabling location tracking. People are repeatedly prompted to grant location access and are told they will get the most out of the app by turning it on. In this environment, “voluntary” disclosure becomes a legal fiction, masking the reality that meaningful participation in digital life often depends on acquiescing to pervasive data collection. Second, the third-party doctrine emerged from 1970s-era cases involving pen registers (Smith v. Maryland) and bank records (United States v. Miller), forms of information that were neither as deeply revealing nor as inseparable from daily life as modern cellphone data. This highlights a broader structural problem: the third-party doctrine has not kept pace with the realities of digital life (see here, here, here, here, here, and here). The doctrine’s underlying assumptions about limited exposure and genuine choice fit uneasily with a world in which participation in social, economic, and civic life increasingly depends on digital services. Seen in this light, the case exposes a deeper tension: whether the Fourth Amendment doctrine will adapt to technologies that make constant data generation functionally unavoidable.

Revisiting Carpenter

At the heart of both the government’s and the petitioner’s claims lies the impact of Carpenter v. United States on Fourth Amendment jurisprudence. The government attempted to confine Carpenter to its facts, portraying it as a narrow decision limited to the long-term collection of historical cell-site location information. On this view, Carpenter is primarily about the duration of surveillance, and therefore does not extend to the shorter time frame involved in this geofence warrant.

This restrictive reading resembles the “mosaic theory” of privacy, which has been widely criticized, even by the scholar most prominently associated with it, Orin S. Kerr. It suggests that privacy concerns arise primarily when the government aggregates large quantities of data over time. However, as the petitioner implicitly argued and as emphasized in the amicus brief filed by Privacy & Surveillance Accountability, Inc., Carpenter is not solely about the duration of surveillance. Rather, it focuses more broadly on the depth, breadth, and revealing nature of the data the government can obtain about an individual, not merely the length of time over which it is collected. In this sense, Carpenter marks a departure from the traditional place- or thing-based focus of SCOTUS Amendment jurisprudence. The decision signals a new kind of expectation of privacy inquiry, one that asks how much the government can learn about a person, regardless of the physical source of the information.

This shift is reinforced by the Court’s use of a contextual, totality-of-the-circumstances analysis. Rather than treating disclosure to a third party as dispositive, Carpenter treats it as only one factor among many in assessing reasonableness. A transformative reading of Carpenter, therefore, would significantly narrow the scope of the third-party doctrine in the digital age and subject large-scale, suspicionless data collection practices such as geofence warrants to meaningful Fourth Amendment scrutiny.

At the same time, even this broader interpretation has limits. The Court emphasized the unique nature of digital data and suggested that heightened protection may be warranted where surveillance reveals the “privacies of life” or involves technologies that have become indispensable to modern existence. Thus, Carpenter may be best understood not as a wholesale abandonment of prior doctrine, but as an equilibrium-adjustment, one that recognizes that certain digital technologies are so pervasive and intrusive that traditional assumptions about voluntariness and exposure no longer hold.

Conclusion

At Stake in Chatrie is whether the Fourth Amendment can meaningfully regulate suspicionless, technology-driven data sweeps that reveal detailed information about where people go, whom they associate with, and what spaces they occupy. A decision that treats geofence warrants as outside the scope of the Fourth Amendment would reinforce the continued dominance of the third-party doctrine and risk normalizing broad digital dragnets as routine tools of investigation.

By contrast, a ruling that views geofence warrants through the lens of Carpenter could mark an important step in adapting constitutional protections to the realities of the digital age. Recognizing that large-scale location data exposes the “privacies of life” would place principled limits on generalized searches and acknowledge that formal notions of consent do not fully capture how modern technologies operate in practice.

The implications would also extend to data privacy standards, everyday law enforcement practices, and the ways in which technology companies collect, store, and respond to requests for user location data. The Court’s reasoning may influence how lower Courts assess other emerging “data scan” techniques, including reverse keyword searches, cell tower dumps, artificial intelligence–driven queries, and other large-scale, algorithmic, or reverse-search methods that rely on bulk data held by private companies.

The Question of Delay in the SC’s Delhi Riots Bail Order [Guest Post]

February 19, 2026 Gautam BhatiaLeave a comment

[This is a guest post by Rehan Mathur.]

The SC passed a detailed order in the bail applications of several accused in the 2020 Delhi Riots case on 05th January 2026. As per the order, the SC granted bail to all but two of the accused; Umar Khalid and Sharjeel Imam. Throughout the 144-page order, the SC has relied upon the distinction between the central role of Khalid and Imam with the incidental role of the other five co-accused as the basis to deny bail to them. While several commentators have already said much about the order (see here and here), the piece attempts to unpack the order’s rationale by examining how the order determines delay attributable to the accused and its consideration of the prima facie case and role of the accused as relevant factors for determining bail pleas premised on delay and prolonged incarceration. It concludes that the order worsens the jurisprudence of bail in UAPA cases.

Article 21 & Prolonged Incarceration

The threshold for the grant of bail in UAPA Cases is extremely high courtesy Section 43D(5) of the UAPA and the SC’s decision in Zahoor Ahmad Shah Watali. Applications of bail, require the bail deciding judge to look only at the chargesheet and case diary to determine if the accusation against such person is prima facie true. If the accusation is prima facie true, bail cannot be granted. However, the Supreme Court recognised a completely new basis for grant of bail in UAPA cases in K.A. Najeeb. In Najeeb, the Court held that the presence of statutory restrictions under Section 43(D)(5) could not oust the ability of constitutional Courts to grant bail in cases to protect rights under Article 21 where the likelihood of trial being completed within a reasonable time was slim, and the period of incarceration already undergone was substantial.

It is important to note that Najeeb was decided by a bench of three judges, and is binding on smaller two judge benches. But this is completely ignored in the 5th January Order, which goes on to whittle down the import of Najeeb. The Court does this by arguing that the delay caused was attributable to the accused, and by the introduction of new considerations for grant of bail under this ground.

Delay Attributable to the Accused

The Order clarifies that the inordinate delay which forms the basis of grant of bail should not be attributable to the accused. The Court notes that several objections, deferment requests and sequencing issues were raised on behalf of the accused before the trial court. Particularly, the Court also mentions delay during the stage of disclosure under Section 207 CrPC [Para 38]: “At the stage of compliance under Section 207 CrPC itself, the Trial Court noted that despite repeated directions, certain accused declined to receive copies of the charge-sheet in the manner directed, insisted on alternate modes of supply, or filed successive applications, necessitating further procedural orders and contributing to delay at the pre-charge stage. These aspects emerge from the record and are not matters of conjecture.“

This analysis by the Court is faulty on two grounds. First, the right of an accused to receive documents relied upon by the prosecution has been recognised as a fair trial right under Article 21. Therefore, it follows, that to meaningfully exercise this right, the accused may file applications to seek supply of illegible, missing or unrelied-upon documents. The SC in Manish Sisodia rejected this exact argument of the ED by holding that in cases of voluminous records, the accused must be given a reasonable amount of time to inspect documents. In the present case, the record extended up to nearly 22,000 pages. The time taken by the accused during the 207 stage to meaningfully exercise their fair trial right cannot thus be equated with delay attributable to the accused to defeat the grant of bail.

Second, even assuming that the accused did contribute to the delay at the 207 stage, the order makes no mention of who caused this delay. In FIR No. 59 of 2020, there are 20 accused persons. The Court makes no analysis on which of these 20 accused persons caused the delay. It could be possible that none of the seven accused seeking bail before the Court had made any contribution to the delay. Would it then be reasonable to deny bail to Accused X, simply because Accused Y might have contributed to some delay? Such an analysis is entirely missing from the Order. The Court should have undertaken an individualised analysis of the delay caused by each accused rather than attributing delay to all of them. Despite the order placing so much emphasis on the individual role of each accused, it is ironic that the order failed to employ such an approach where it was most needed.

Mixing of two standards

After clarifying the scope of delay, the Court considered the four relevant considerations for the grant of bail under Article 21 [Para 56]:

The outcome of such scrutiny must be determined by a proportional and contextual balancing of legally relevant considerations, including (i) the gravity and statutory character of the offence alleged, (ii) the role attributed to the accused within the alleged design or conspiracy, (iii) the strength of the prima facie case as it emerges at the limited threshold contemplated under the special statute, and (iv) the extent to which continued incarceration, viewed cumulatively in the facts of the case, has become demonstrably disproportionate so as to offend the guarantee of personal liberty under Article 21.

While the Supreme Court itself has negated the applicability of the gravity of the offence in deciding bail on the grounds of Article 21, the two other factors, namely, the accused’s role and the strength of the prima facie case are far more problematic.

Addressing the factor of prima facie case first, it is clear that introducing this factor is a clear mixing up of two different grounds. The question of whether a prima facie case is made out or not is anyways a question of the merits under Section 43D(5), which the Court can address. Whether a prima facie case exists or not has nothing to do with why Najeeb and other decisions have recognised such a ground for grant of bail. This ground attempts to remedy the prolonged incarceration of the accused, which if not for the grant of bail amounts to a violation of the Accused person’s Right to liberty and speedy trial under Article 21. Mixing of the two standards amounts to making the accused’s constitutional right subservient to the draconian requirements of a special statute, in complete ignorance of constitutional supremacy.

The Role of the Accused & the Purpose of Bail

The alleged role of the two accused is also the second problematic factor considered by the Court and forms its ultimate premise for denying them bail. This begs the question then, what really is the relevance of the alleged role played by an accused person for the grant of bail under Article 21? The order seems to throw some light on this [Para 100]:

The continued detention of those alleged to be the architects of the conspiracy may be required to safeguard broader security interests and deter future acts, whereas the rationale for continued incarceration of minor participants is comparatively attenuated once the investigative purpose is exhausted. The Court is therefore justified in calibrating its approach, ensuring that the pursuit of security does not eclipse the principle of proportionality.

The order seems to suggest that bail can be denied simply to protect security interests and deter future acts. But the purpose of denial of bail is very different. Deprivation of liberty by denial of bail is justified to ensure attendance, or prevent witness interference or tampering with evidence. Ultimately therefore, the law is settled that the object of bail is to secure the attendance of the accused at the trial. Using deterrence of future acts as a ground (especially in cases where the accused has no past antecedents) would amount to determining guilt and punishing the accused at the stage of bail, and turns the presumption of innocence on its head.

An accused who might have a tangential role may be highly influential with the ability to influence witnesses, indicating that role in an offence as such has nothing to do with bail. In any case, past antecedents, ability to influence witnesses, or risk of flight are objectively ascertainable, but the role of the accused is something that the Prosecution, and the State has control over. Allowing the role in the crime to be the basis for grant of bail allows States full control over who to make a case against, and prevent those it ascribes an ‘architectural role’ from grant of bail.

It must also be noted that the SC has granted bail under Article 21 in cases where the accused had a major role as per the Prosecution. In Najeeb, the accused was alleged to be one of the main ‘conspirators’ in the crime (who had absconded) yet, bail was granted. Similarly, in Umarmia, the Supreme Court granted bail to the alleged ‘mastermind’ who was charged under various provisions of the TADA amongst others for smuggling a large quantity of arms and ammunition.

While proponents of this position may seek to rely on Shaheen Welfare Assn. where the Court created 4 categories of undertrials on the basis of their role, such as by directing leniency for those who were roped in by virtue of Section 120-B or 147 IPC not being directly involved in any terrorist or disruptive act and those who were found possessing incriminating material. While the categorisations of the judgement may themselves be questionable, even the judgement notes that those who are directly involved in terrorist acts under Section 3 TADA, may be entitiled to grant of bail if they have been in prison for five years or more, and their trial is not likely to be completed within the next six months. The only exception to this rule is when the Court concludes that the accused’s antecedents are such that releasing them may be harmful to the complainant, their family or to witnesses.

This signifies that even the dictum of Shaheen Welfare Assn. permits an alleged ‘mastermind’ who has committed terrorist acts to be granted bail under Article 21, in cases of prolonged custody. This follows because the violation of Article 21 rights arising from prolonged incarceration with no likelihood of completion to trial affects all accused equally irrespective of the role or nature of their involvement in the alleged prosecution case. Thus, the SC order creates an irrelevant test based on the role of the accused to selectively decide who should be granted bail without any analysis of whether the accused will actually tamper with evidence, influence witnesses or abscond from trial.

Some Concluding Thoughts

The Order allows Umar Khalid and Sharjeel Imam to seek bail again after the expiry of a period of one year from the date of the order, or upon completion of examination of the protected witnesses relied upon by the prosecution. This direction reflects the approach of the Court when it in an earlier portion of the judgement notes [Para 50]:

The Court must also bear in mind that it is not confined to a binary choice between continued custody and unconditional release.Where delay becomes a matter of constitutional concern, appropriate directions for expeditious trial, prioritisation of witnesses, or periodic review of progress may be issued. Such measures are constitutionally significant responses that address the vice of delay while respecting the statutory framework. The liberty to renew a prayer for bail upon continued stagnation may also be preserved.

The issue with such directions is that they fundamentally miss the point of invoking Article 21 for seeking bail. An accused’s right under Article 21 is violated due to prolonged incarceration already suffered, and if there is no likelihood of trial finishing. Looking at only the latter without any consideration for the prolonged incarceration, leaves the accused at the mercy of the Prosecution and the Trial Court. Measures for expediting trials, prioritising witnesses may be relevant but cannot completely remedy the suffering caused by prolonged incarceration, since these directions may not actually result in movement, let alone completion of the trial. In the present case, the order provides for a direction for renewal of bail. But charges are yet to be framed in the present case, and thus, it is extremely unlikely that the examination of protected witnesses will take place. Why then should the accused suffer detention for an additional year? There is no answer.

The Order, thus adds to the already problematic jurisprudence of bail in UAPA cases, like Watali and Gurvinder Singh by mixing two independent grounds, by penalising the accused for exercising their fair trial rights and giving the state full authority to ascribe anyone with an ‘architectural role’ in an offence to deny them bail. Whether it will turn out to be the outlier, or the norm in the sea of inconsistent decisions on bail in UAPA, only time will tell.

Re-Viewing Review Powers of the Supreme Court [Guest Post]

February 19, 2026 Gautam BhatiaLeave a comment

[This is a guest post by Kunsh Jain.]

Article 137 allows the Supreme Court (‘SC’) to review its own judgements. This review power is subject to, in the absence of parliamentary legislation, Order XLVII of the SC Rules, 2013 (‘Rules’). The SC has highlighted a worrying trend of overruling rulings since several orders have been recalled or modified by review petitions. This post attempts to uncover the contours of invoking these review powers. This is done firstly, by critiquing how post the Kantaru review petition the review jurisprudence trend shifted from restrictive to expansive, and secondly, analysing 4 grounds, namely: the ‘eroding plausible-impossible distinction’, ‘bench shopping’, ‘commercial implact;, and ‘not following judgements’ as questionable expansion of review grounds.

A review is sought to exercise corrective powers post a final order by the SC. It does not amount to re-hearing a case but is to be used sparingly to not let it become an intra-appellate mechanism. Its function is to rectify a patent error in a final judgment, not to revisit matters conclusively decided. The dilution of this distinction erodes Article 141’s promise of finality. This finality is central to the rule of law and as a case travels up the vertical hierarchy of courts, the jurisdiction of the court gets narrower. Thus, “a review cannot be allowed to be an appeal in disguise”.

The SC had restricted review jurisdiction starting from Northern India Caterers (1979) to Kamlesh Verma v. Mayawati (2013) (‘Mayawati’) justified only when there exist substantial and compelling reasons to do so. However, these reason and their precise standards are unclear in recent jurisprudence. Review petitioners usually try to ‘force-fit’ reasons into the limited grounds considered for review. Rule 1 Order XLVII of the SC Rules, 2013 was supposed to restrict this otherwise unfettered power to admit review, but post the Kantaru Review in 2020, a pandora’s box of challenging final judgements via review petitions opened.

Jurisprudence expanding Jurisdiction

In Kantaru Rajeevaru v. Indian Young Lawyers’ Association (‘Kantaru’), there were several review petitions after the SC permitted women of menstruating age into the Sabarimala temple. By a narrow 3-2 split, these reviews were maintained and referred the review to a larger bench. The dissent rightly attempted to restrict the review jurisdiction arguing that the grounds under Rule 1 Order XLVII of the rules were not met.

However, the 9 Judge Bench (‘9JB’) admitted the Sabarimala review petition with a flawed reasoning. It escaped the set standards with a disjunctive interpretation, implying an unfettered power to review as the general rule, and Rule I applied only to civil and criminal cases. Hence writs were made reviewable without having to meet an ‘error apparent’ standard. This author critiquing this had – accurately, as it turned out – predicted that in future, judges would not provide adequate reasoning for exercising review jurisdiction in any proceedings other than civil or criminal ones and this would turn a “reluctant resort” into “whims and fancies”.

This artificial distinction of writs from civil and criminal proceedings largely set the trend for the unfettered review petitions as we see them today. It was a recognised rule that a review standard followed the CPC’s limited grounds of i) discovery of new evidence, ii) error apparent on the face of record and iii) any such reason which of an analogous nature, and in a criminal proceeding, only upon an error apparent. By distinguishing writs, these standards no longer bind the Court, leading to multiple unsubstantiated review petitions after every lost SC case.

Judgements do not even always follow the Kantaru idea of a writ versus a civil-criminal proceeding binary. For example, the recent review in CREDAI v. Vanashakti (‘Vanashakti’) had a civil appeal clubbed with writ petitions, or in Mayawati where civil irregularities led to a criminal inquiry against which a writ was filed. The governing rules of review for such ‘mixed’ proceedings are lost. Even beyond the expansive powers for writs, the rationale for exercising review powers in civil and criminal proceedings is uncertain. The scope of an ‘error apparent on the face of record’ or ‘any other reason’ being analogous being noscitur a sociis to the first two conditions was sought to be clarified in Mayawati. However recent deviation makes the scope of Rule 1 unclear.

A two-judge bench in Shikhar Chemicals Versus State of UP recently recalled its order directing stripping the criminal roster of an Allahabad HC judge. Despite it being a criminal SLP, it was reviewed by an “undated letter by the CJI”. On procedure, per Govt of NCT Delhi v. Rathi Steels, the person applying for review (CJI herein) must demonstrate that he is a person aggrieved (unclear how so). On substance, the only ground for reviewing criminal proceedings is of an error apparent. What constituted such an error in the now-deleted Paras 25 and 26 that issued roster directions to the Allahabad HC, is something only the CJI’s letter can answer.

Making Impossible the Plausible

First, review cannot be sought where the court has taken a view, however erroneous, if that view is one of two possible views and there lies no patent error. In the case of Kalyani Transco v. Bhushan Power (‘Bhushan’), the order-in-review undermined commercial confidence by unravelling a resolution plan already invested in. This was due to lack of adherence to procedural requirements and not meeting the timelines that were held critical to the IBC framework. However the extraordinary procedure for seeking review is critiqued here.

Oddly, using a new SLP against an enforcement order rather than a review, the order was stayed. The CJI’s order is titled ‘civil appellate jurisdiction’ and not the inherent/review jurisdiction. Secondly, the order was recalled and reversed, because i) judgements were not correctly followed ii) it was based on incorrect factual aspects and arguments not advanced were considered. These grounds may be used in appeal, however a narrower review standard isn’t sufficiently met by alleging per incuriam. The earlier interpretation of IBC’s procedural requirements was a plausible view, however erroneous that may be, and not an impossible view and hence, no ground for review.

This plausible-impossible distinction was blurred even in Vanashakti where the judgement in review was about the validity of post-facto environmental clearances. The earlier order sought to end this practice of ‘backdoor’ regularisation, but the review bench held by 2-1 majority that this was reviewable, and further reversible due to contravening earlier judgements. However, Justice Bhuyan’s dissent brings to light the problems that review jurisdiction faces currently.

The existence of a well-reasoned view, seeking to bring India in line with environmental jurisprudence and global standards, is overturned for ‘not following judgements’. These earlier judgements had used their exceptional Art. 142 powers to regularise and this ‘relief’ must not bind the court as ‘ratio’. Further, a concern repeated with both Justice Oka in Vanashakti and Justice Trivedi in Bhushan is that once the authoring judge retires, a review is sought and a re-attempt at a favourable decision is made. This furthers the courts polyvocality, distinguishing a recently retired judge from others of the same court. This concern was also raised by Justice Nagarathna warning that judgements were anchored in ink, not sand and cannot be tossed out merely because faces have changed.

Who Decides Who is to Decide

Rule 3, Order XLVII of the Rules states that “An application for review shall as far as practicable be circulated to the same Judge or Bench of Judges that delivered the judgment or order sought to be reviewed”. Changing the composition of benches by admitting review post-retirement, or diluting the original bench strength by adding additional judges goes against the intent of the rules where if the error was so apparent, the same judges could re-look and correct their error. This makes review, a ground for bench-hunting to obtain a favourable decision. Often in reviews with changed bench composition, the judge from the earlier bench maintains the original view taken, however the newer judge(s) give a different view.

In Vanashakti, the 2JB’s order by Oka and Bhuyan, JJ was reviewed post Justice Oka’s retirement by a 3JB composed of CJI Gavai, Chandran and Bhuyan, JJ. This article highlights the reversal via a miscellaneous petition being before a 3JB being a problematic substitute to review. This 3JB gave a split verdict of 2-1, with Justice Bhuyan maintaining the original view and calling the review a step in retrogression.

Elsewhere, in a property dispute between ISCKON Bangalore and ISCKON Mumbai, Oka and Masih, JJ had ruled in favour of the Bangalore chapter. Upon a review petition, Justice Maheshwari replaced retd. Justice Oka’s and admitted the review condoning the delay, but Justice Masih held that there existed no error apparent and hence grounds for review were not met. This split decision led to another completely new three judge bench issuing notice and leaving the question on review jurisdiction for open for the past three months remarking “What Lord Krishna might be thinking of all this”.

This also confers a special power in the hands of the CJI as the ‘master of the roster’. If there exists a split-decision, as seen in the ISCKON dispute and even in the Akola riots case, the review is referred to the CJI for further directions. With the power of constituting benches, there exists a re-view of the original order that can be created with a change in the bench. Even in civil review, Rule 5 Order XLVII of the CPC precludes other judges from hearing review highlighting the intent of self-correction by the original judge(s). It creates an exception for retiring/transferring judges, however, even in rare cases reviewed by the same bench, the grounds for Review are not followed.

Rule 2 mandates filing review petitions within 30 days of a judgement/order placing primary intent of swift finality after long appeals, granting condonation of delays in filing review petitions is common practice. In NHAI v. Tarsem Singh, the same bench admitted a review petition 9 months later, on the grounds highlighted that had far-reaching implications of Rs. 32,000 crores, and not Rs. 100 crores as mentioned in the earlier judgement. While this could be fit under the ground of “new material/evidence” it raises another important question: Can commercial impact be a ground for review under Rule 1, Order XLVII of The Rules?

A Reason not Reasonable enough

Every judgement will have certain commercial and political implications to the disadvantage of key stakeholders. The presumption is that the judges are cognizant of the same while delivering the judgement. Bringing these up in a review petition relies on post-judgement public sentiment and politico-economic considerations to reverse the effect of an final judgement, thereby undermining the judicial process followed by it. With no set standard of seeking review by the impact created, there exists confusing jurisprudence on whether this can be fit into the ‘any other reason” ground of seeking review.

In Vanashakti, the arguments accepted in favour of review were of public constructions already finished and their demolishing resulting in a loss of public money. Examples of an AIIMS hospital and Greenfield airport built on taxpayers money being demolished as an the effect of the 2JB judgement constituted a ground for review. However, the implications also extended to a larger number of private players and projects subsequently benefitting from the revised judgement. Even in Bhushan, the judgement in review was criticised for undermining commercial certainty and investor confidence and its economic impact certainly played some role in admitting that petition.

But the long-term consequences that these judgements seek to protect are equally crucial too. In Vanashakti, the 2JB had sought to alleviate the pollution issue by tuning India’s environmental jurisprudence to global standards. They had relied on impact assessment studies considering public stakeholders in the environmental discourse and concluded noting the Delhi smog as a visible impact. This acknowledged impact should not be then backtracked upon using a review petition filed by “builders and developers who had shown scant regard for the law”. What then, is the distinction between impacts that make a case for review and impacts that don’t?

In State of West Bengal v. Baishakhi Bhattacharya (‘WB teachers’), the court rejected a review petition to its judgement invalidating 24,000 appointments by the WB School Service Commission from 9 years back calling it a “re-hearing on merits” and hence not permissible. It “was fully conscious of the heartburn and anguish caused by the invalidation of the untainted appointments” but the matter did not deserve a review petition as all aspects were considered.

Hence, the result is a skewed jurisprudence in which commercial stakes routinely justify review, but the far-reaching effects on ordinary citizens do not. This reveals how “impact” has become an incoherent and inconsistently applied ground under the “any other reason” of Rule 1, Order XLVII.

Error Apparent or Apparent Expansion

Another equally popular ground of seeking and admitting review is by “error apparent” of not following the law as laid down in previous judgements. To answer whether ‘not following earlier judgements’ constitutes error apparent, the review jurisprudence pre-Kantaru must be looked at. In Mayawati, the SC clearly lists when a review will not be maintainable. It held that older arguments cannot come up in a review petition as to re-hear a case, and an error that needs to be fished out and searched is not an error apparent. Thus, if a judgement was engaged with and/or distinguished from, then per Mayawati a review petition must not exercise appellate-like powers to reverse a final judgement.

However, the review judgement in Vanashakti had an interesting section on judicial discipline and propriety. Citing a plethora of judgements on how stare decisis is integral and following precedent is key to judicial discipline, the 2JB’s judgement was overturned as it had not followed certain paragraphs of judgements “that could have persuaded it to take a different view”. The effect of Vanashakti is that if a precedent was followed largely but differed in an interpretation of certain paragraphs, the judgement is per incuriam and hence, constitutes a sufficient ground for review. The reason cited is that “predictability and certainty is important…. And discipline is a sine-quo-non for effective functioning of the judicial system”.

It is the SC’s liberty to apply the general rules to specific facts of each particular case to obtain a just outcome for the parties and set precedent for the country. The order-in-review in Vanashakti had engaged with said judgements and held that it couldn’t be bound by the special Art. 142 relief granted and focused on the legal principle that post-facto ECs were alien to environmental jurisprudence. This is not judicial indiscipline or distinguishing by “minor difference in facts”, but following the SC’s prerogative of setting progressive law. By refusing to rely on environmentally regressive reliefs granted in earlier judgements, it had sought to not shy away from a principle’s application, as this post argues it previously has.

Allowing “not following judgements” to constitute a review ground takes away from transformative judicial interpretation. Rupa Ashok Hurra v. Ashok Hurra, which engaged with these questions of discipline and certainty sufficiently, had highlighted the narrow ladder moving from appeals-review-curative jurisdiction. Mayawati had laid the ground for when a review is not maintainable. However, review petitons and admissions post Kantaru have negated this restrictive sense of review powers and made it into what Krishna Iyer, J had warned be, a shot at the moon.

Conclusion: Restricting Review Powers

Review powers of the SC remain in an unchartered territory because of its unfettered power subject only to the parliament or the rules made by the SC itself. The parliament will not interfere with judicial procedure for fear of being struck down as infringing on the separation of powers even if constitutionally empowered under article 137. Further, why would the UOI who themselves get another chance at “appealing” an unfavourable order, and recently have through another gateway using a presidential reference, choose to limit themselves? Political ramifications and public sentiment are now a ground for review as seen in In re: Stray Dogs and In re: Assent to Bills and the largest litigant does not want to limit the same.

The Rules after Kantaru’s distinction are also interpreted differently to decide a review on its current impact basis rather than a limited set standard. Firstly, the courts can always resort to their inherent powers and Rule 6, Order LV and hold that the Rules do not apply while considering review petitions. Notwithstanding discretionary powers for justice, there must be a clear standard for review and it must be guided by a restrictive reading of Order XLVII of the Rules. This is critical to maintaining the limited jurisdiction as a case travels up the hierarchy and the finality it offers.

Secondly, admissions like the 9JB in Kantaru and the 2-1 majority in Vanashakti dangerously expand the scope of review. These benches themselves set the scope of powers they can exercise and guide future smaller benches. In these scenarios, the dissents, by Justice Nariman in admitting the review petition prior to the 9JB reference, and by Justice Bhuyan in Vanashakti, guide rightly towards a restrictive review jurisdiction amidst increasing pendency, uncertainty and a litigant hoping for finality.

Review powers then, are an intricate balancing act and the power of preventing them from becoming an intra-appellate stage rests with either the parliamentary law or the SC rules. In the absence of a conclusive set standard, it is the judiciary’s prerogative, split into benches but speaking as one, to shape jurisprudence in a way that views review as a careful and controlled power exercised only upon a serious miscarriage of justice.

Surveillance by design: The Unconstitutionality of Section 7(i) of the Data Protection Act [Guest Post]

February 19, 2026February 19, 2026 Gautam Bhatia1 Comment

[This is a guest post by Chytanya S. Agarwal.]

Section 7(i) of the Digital Personal Data Protection Act, 2023 (‘DPDPA’ or ‘the Act’), read with Sections 2(d) and 4(1)(b), allows employers to non-consensually ‘process’ personal data of employees for ‘purposes of employment’ or for preventing ‘liability or loss’ to the employer. Section 7(i) reads as follows:

7. Certain legitimate uses: A Data Fiduciary may process personal data of a Data Principal for any of following uses, namely:— … (i) for the purposes of employment or those related to safeguarding the employer from loss or liability, such as prevention of corporate espionage, maintenance of confidentiality of trade secrets, intellectual property, classified information or provision of any service or benefit sought by a Data Principal who is an employee.

The provision inexhaustively lists corporate espionage and breach of confidentiality as instances of ‘liability or loss’ to the employer. However, it provides no guidance as to what the phrase ‘purposes of employment’ entails. Once these undefined grounds are triggered, employers are empowered to non-consensually ‘process’ employees’ data without any constrains. ‘Processing’, defined widely under Section 2(x), includes surveillance, use, and dissemination of personal data.

In this blog post, I question the constitutionality of this blanket exemption for employer-led non-consensual data processing based on the ambiguous drafting of Section 7(i). My argument is three-fold: firstly, Section 7(i) violates the right to privacy by failing Puttaswamy (Aadhar-5J)’s test of proportionality; secondly, Section 7(i) is hit by the manifest arbitrariness doctrine under Article 14; and thirdly, furthering the Article 14 argument, I argue that the DPDPA itself does not envisage restrictions on Section 7(i) even by way of subordinate legislation. The absence of clear rule-making provisions for Section 7(i) implies that any delegated legislation restricting the exemption would be ultra vires the Parent Act. Thus, the Act, by design, vests employers with unrestrained surveillance powers, strengthening the Article 14 argument.

Right to Privacy

Per KS Puttaswamy (Privacy-9J) [(2017) 10 SCC 1, ¶320-326], the right to privacy emanates from the guarantee of life and personal liberty under Articles 19 and 21. Privacy is a state of freedom from intrusion, with an “effective guarantee of a zone of internal freedom” and cannot cease in public sphere. It is the springboard for enjoyment of all Article 19(1) freedoms (Puttaswamy-9J, ¶402-3, ¶412, Bobde J.). Surveillance per se constitutes a restriction on right to privacy as per Justice Subba Rao’s minority opinion in Kharak Singh v. State of U.P., which was upheld by Puttaswamy (9J). Thus, employer-led surveillance under Section 7(i) is ipso facto an Article 19/21 restriction. This prima facie restriction dislodges the presumption of constitutionality in favour of Section 7(i).

In Puttaswamy cases, the right to privacy was claimed against the State. In contrast, surveillance under Section 7(i) predominantly concerns private employers, in addition to instrumentalities of the State. This issue of claiming right to privacy against private actors is no longer res integra in light of Kaushal Kishor [(2023) 4 SCC 1]. The majority in Kaushal Kishor (¶83) held that rights under Articles 19 and 21 are enforceable even against private persons. In other words, the State has a positive duty to protect rights under Articles 19 and 21 from threats by non-State actors. Despite the conceptual errors in Kaushal Kishore, I invoke this positive duty of the State and argue that the State failed to perform its duty insofar as Section 7(i) gives a blanket and unlimited exemption to private employers for restricting the privacy of employees. Per Puttaswamy (5J) [(2019) 1 SCC 1], any restriction on the right to privacy must pass the test of proportionality. I present a bare-bones prong-wise application of the proportionality test.

Legitimate Aim: This requires that the rights-restrictive measure must be traceable to reasonable restrictions in Part III. Per Kaushal Kishore, restrictions on Article 19(1)(a) enumerated in Article 19(2) are exhaustive and additional grounds of restriction cannot be imported to curtail 19(1)(a) rights. “Purposes of employment” do not find any remote mention within 19(2) grounds. Nor can it be traced to ‘public order’ or ‘morality’, given their narrow interpretation in past judgements [see Anuradha Bhasin v. Union of India, (2020) 3 SCC 637]. However, the restrictions under Article 21 are broad and unenumerated – it can be validly restricted by the procedure established by the law. For the sake of simplicity, we assume arguendo that this prong is fulfilled.

Suitability: This requires that the rights-restrictive measure must have a rational nexus to the object sought to be achieved. That is, the means should reasonably further their objects. This approach is deferential to the rights-restrictive measure. The enquiry is confined to assessment of a relationship between means and ends. So long as the means advance the ends to any degree, the suitability prong is fulfilled. Due to the deferential nature of enquiry, we assume that Section 7(i) fulfils the second prong.

Necessity: The necessity prong, per Modern Dental College (¶53-58), requires the absence of equally effective and less rights-restrictive measures. Puttaswamy-5J tempered the strictness of necessity prong by incorporating Bilchitz’s ‘moderate interpretation of necessity’ test. Instead of being “equally effective”, such alternative measures must realise the object in a “real and substantial manner”. While my enquiry here is not exhaustive, I adduce Section 13 of the Personal Data Protection Bill, 2019, as a feasible alternative. This provision was based on the recommendations of Justice BN Srikrishna Committee Report. Section 13 has the following restrictions:

Non-consensual processing of employee’s personal data excludes sensitive personal data.
This provision comes into effect only if consent of data principal (employee) is not appropriate or involves disproportionate effort on part of data fiduciary (employer).
Purpose Limitation: The purposes for which such data may be processed are limited to 4 situations: recruitment and termination; provision of service/benefit sought by employee; attendance verification; and assessment of performance of employee.

Section 13 of the 2019 Bill substantially fulfils the purpose of Section 7(i) – it provides for monitoring for purposes of assessing performance and protecting confidentiality monitoring. Section 13 of the 2019 Bill is also less rights-restrictive vis-à-vis Section 7(i) for one main reason: the exemption does not extend to sensitive personal data. Thus, Section 7(i) fails the necessity analysis.

It is noteworthy to mention here that in Ramesh Chandra Sharma v State of Uttar Pradesh, the Supreme Court added a fifth prong to the test of proportionality, viz., “adequate safeguards”. However, as argued by Rudraksh Lakra, this prong already pre-exists within the necessity stage of the proportionality test. If we assess Section 7(i) in the light of this, it visibly lacks adequate safeguards on the extent to which employers can non-consensually process employees’ personal data.

Balancing or ‘proportionality stricto sensu’: This final prong requires that the restriction on the right should be balanced with the benefit gained from such restriction. It requires a comparative assessment of the net gain, i.e., whether the positive benefits of the restriction outweigh the negative impact on the right. In this regard, I invoke Schmitz’s economic theory of workplace surveillance. Per Schmitz, privacy laws can be justified even on the basis of non-moral, economic values because they maximise the total surplus produced by an employer and employee. Schmitz employs a principal-agent model where the employer seeks to induce extra effort from the wealth-constrained employee. Because effort is unobservable, employers pay efficiency wages to incentivise unobservable effort. Such incentives are called positive rents. To reduce this rent, employers invest in surveillance technologies. However, the employer’s sole interest is to maximise their own profit and not the total surplus.

Per Schmitz, just to reduce positive rents, an employer might invest in surveillance technology even if the monitoring costs exceed the additional surplus generated by the higher effort level. Here, the employer wastes resources to redistribute wealth from the employee to themselves without any increase in total surplus. Schmitz argues that privacy protection laws at the workplace prevent this “socially wasteful rent-seeking,” thereby increasing the total surplus. Moreover, constant surveillance causes several externalities which impact employees’ productivity (by causing stress, anxiety, depression, and musculoskeletal problems) and increases health costs. Applying Schmitz’s theory to Section 7(i), privacy protection arguably will cause a greater net gain to the society vis-à-vis no privacy protection at the workplace. This makes the blanket exemption under Section 7(i) socially inefficient. Since the net gain under privacy protection exceeds the net gain under no privacy protection (which is the case with Section 7(i) exemption), Section 7(i) fails to pass the balancing test.

Article 14

Per Shayara Bano, manifest arbitrariness denotes legislative acts which are capricious and lack a determining principle. Per KA Abbas, if a rights-restrictive law allows its administrators to exercise unrestrained powers, it must be held unconstitutional.

A quick look at Section 7(i) of DPDPA shows that the employer can process personal data of employees non-consensually without any limits. Apart from Sections 11 and 12, which give minimal rights to the Data Principal (here, the employee), the Data Fiduciary (here, the employer) is vested with unlimited powers to surveil and process the personal data of employees.

Moreover, per Shreya Singhal, expressions of inexactitude, that are so broad as to cover any subject covered by it, are unconstitutionally vague and arbitrary. If the definition of the word offers no guidance, it would create unfettered discretion. The grounds for data processing provided under Section 7(i) are hit by vagueness, ambiguity, and overbreadth due to the following two reasons. Firstly, the import of “purposes of employment” is unclear. Unlike the Justice Srikrishna Committee Report which defined it to include 4 grounds (recruitment and termination; provision of service/benefit sought by employee; attendance verification; and assessment of performance), the DPDPA leaves its scope undefined. This creates ambiguity regarding the extent of permissible surveillance. Similarly, term ‘corporate espionage’ is undefined. Would surveillance of employees outside the workplace be necessary to prevent corporate espionage? If yes, the scope of surveillance would barely have any limits due to functional creep. Secondly, the list provided under Section 7(i) is inexhaustive and inclusive. This is clear because the usage of “such as” implies that the list is only indicative. This leads to a situation wherein the employer can claim grounds of surveillance not enumerated in Section 7(i) so long as they have some nexus (howsoever remote or tenuous) to employment or prevention of “loss or liability”.

III. On the possibility of restraints via subordinate legislation

The DPDP Rules, 2025, do not create any limits on Section 7(i) exemption. Interestingly, the Central Government has no rule-making powers to make a subordinate legislation to define the limits of Section 7(i). The provision conferring rule-making powers, Section 40, only mentions clause (b) of Section 7, with no mention of clause (i) of Section 7.

Moreover, any rule defining the limits of Section 7 cannot be made under the general rule-making provision under Section 40(1). As held in Global Energy Ltd v. Central Electricity Regulatory Commission, rule-making power “for carrying out the purpose” of the statute is a mere general delegation and no guidelines can be laid down thereunder. Similarly, per Sukhdev Singh v. Bhagatram, the Central Government cannot travel beyond the rule-making provision to make rules or guidelines. So any rules or guidelines framed by the Central Government to limit the exemption under Section 7(i) would be ultra vires the parent Act.

In short, the scheme of DPDPA itself does not envisage any limits on the extent of surveillance permitted under Section 7(i). Nor does it empower the Central Government to make rules in this regard. This strengthens the Article 14 argument – Section 7(i) of DPDPA, by not envisaging limits on the exercise of rights-restrictive powers, is hit by the doctrine of manifest arbitrariness.

IV. Conclusion

The only way to preserve the constitutionality of Section 7(i) seems to be a judicial reading down of the provision. For this, three limits can be read into Section 7(i):

Principle of Purpose Limitation: To address the vagueness and overbreadth inherent in the phrase “purposes of employment,” the scope of Section 7(i) must be strictly construed. Drawing on the Justice Srikrishna Committee recommendations, the term should be read to exclusively include only those purposes that are essential to the employment relationship: i.e., recruitment, termination, provision of sought services or benefits, attendance verification, and the assessment of performance. Limiting these grounds prevents the functional creep of surveillance where employers might otherwise claim a remote nexus to employment to justify invasive monitoring of an employee’s private life.
Safeguards for Sensitive Personal Data: Section 7(i) fails to distinguish between general and sensitive personal data. To pass the necessity prong of the proportionality test, Section 7(i) must be read to strictly exclude the non-consensual processing of sensitive personal data. This includes biometric/health data, monitoring of private communications, etc., which should only be processed with the informed consent of the employee, rather than being subsumed under a blanket exemption.
Spatial, and Temporal Limits: Surveillance must be limited to the performance of actual work tasks and should not intrude into the employee’s private sphere, especially during telework or outside of official working hours. Illustratively, the tests of “in the course of employment” and “arising out of employment” in labour law can be useful for determining the spatial and temporal limits of surveillance.

Between Constitutional Fidelity and Political Demand: A Critique of the Kenyan Court of Appeal’s CDF Decision – II [Guest Post]

February 19, 2026February 19, 2026 Gautam Bhatia1 Comment

[This is a guest post by Kenson Mutethia.]

From an objective reading of the Court of Appeal’s NGCDF Act Judgment it is fair to state that as the Supreme Court prepares to confront the second round of the NGCDF challenge, it will not only be deciding on doctrine but also confronting competing constitutional narratives. This is because the real problem with NGCDF is not merely MPs’ appetite for control over resources but also voter expectations. Specifically, many MPs genuinely believe that the only contributions voters recognise are visible, localised projects such as graded roads, classrooms built, bursaries issued and water points commissioned. Ironically, many MPs themselves seem to internalise the idea that representation, legislation and oversight are inconsequential. We recently heard the Speaker urge MPs to remember at least the laws they pass. If to some of us this sounded like just another sound bar or a light moment, it should be even sadder, because that moment captured something deeper about how lightly our very own lawmakers treat the legislative function.

While under the Constitution, (i) Parliament’s role is to legislate, represent, and oversee, and (ii) project implementation falls within the Executive’s purview (at both the National and County levels) to avoid blurring the separation of powers, the “lived political reality” is that in the Kenyan political environment, legislative work, oversight and representation feel abstract and politically unrewarding. This “lived political reality” would explain why, in the NGCDF dispute, the National Assembly argues that the NGCDF establishes a lawful framework for the decentralisation of national government services. To it, as accepted by the Court of Appeal now, the constituency serves as an administrative convenience through which national functions are delivered more closely to the people. The adversaries respond that what is described as decentralisation is, in principle and in effect, the creation of a parallel, constitutionally unknown third tier of government. Yet a constituency is not a recognised unit of service delivery under the Constitution; rather, it is an electoral unit. To them, converting it into a funding and implementation unit risks distorting the architecture of devolution and the overall constitutional integrity framework.

However, in its judgment, the Court of Appeal appeared to accommodate the “lived political reality”. To achieve this, the Court of Appeal: (i) suggested that the basic structure doctrine, which might cast doubt on the transformation of constituencies into service-delivery units, does not apply in Kenya, (ii) downplayed concerns about duplication of funding, confusion in project implementation and uncertainty about which level of government is responsible for particular projects. To obscure the deference given to the “lived reality”, however, the court resorted to legal sophistry and judicial parlance (legal sophistry is a ‘clever’ approach that entails making superficially sound but disingenuous and deeply flawed decisions). Specifically, the Court offered three key responses to the issues raised against NGCDF. First, claims of confusion during project implementation are unsupported by reality, as the NGCDF has been operational since 2016. Second, any disputes can be resolved through established intergovernmental mechanisms. Third, and most importantly for the Court, the NGCDF is “clearly” limited to national government functions (for a critique on this blog, see here).

The most fascinating bit in the judgment, however, is how the Court of Appeal found fault primarily with section 43 (9) of the NGCDF, which tethers the independence and tenure of the Fund Account Manager to the political tenure of the Member of the National Assembly (holding that the linkage clearly entrenches political influence). Yet (rather, curiously), the Court did not see comparable difficulty with parliamentary involvement in the Fund’s governance, particularly in approving the appointments of those who administer it. Is it not too fantastic for the Court to assume that parliamentary control was sought only through section 43 (9)? Did the Court deliberately overlook that it was equally plausible that approval powers over key administrators function as a subtler but equally potent mechanism of control? If Parliament participates in approving those who run the Fund, while also retaining political visibility and influence at the constituency level, the separation between legislation and execution becomes porous. The concern quickly metamorphoses from tenure linkage to structural proximity. And we know, perhaps too well, if it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck.

As the Supreme Court revisits the issue, it may need to ask more difficult questions than the Court of Appeal did. Questions beyond the mere rhetoric of whether disputes have arisen since 2016. It is a trite constitutional principle, as established by the Supreme Court in the first round CDF Judgment, that the longevity of a practice does not confer constitutional legitimacy. Second, the fact that confusion has been (or could be) manageable does not mean the design is sound. Third and most importantly, limiting the Fund to “national functions” does not, by itself, prevent overlap in practice.

Notably, the Supreme Court has already endorsed in Law Society of Kenya v Attorney General & another at para 40 the purpose and effect test (also known as the “object and effect test”) as a standard method for resolving disputes concerning the constitutionality of statutes and the infringement of human rights and fundamental freedoms. With this test, a court asks the appropriate questions, which usually result in a conclusion that a law with a facially benign object or purpose is unconstitutional if its effects (intended or unintended) are unconstitutional. The purpose-and-effect test is especially useful in democracies such as Kenya, where constitutional challenges are more practical than theoretical. In fact, this appreciation of the underlying context of high-stakes constitutional challenges (an approach I loosely refer to here as “constitutional realism”) might have informed the landmark BBI decision to preclude the President and other State Officers from serving as promoters/initiators of the “popular initiative” amendment process under Article 257 of the Constitution. To illustrate, on paper, there is nothing sinister about the President moving constitutional amendments through popular initiative; after all, it is often said that “…the President is the only national voice in affairs. Let him once win the admiration and confidence of the country, and no other single force can withstand him; no combination of forces will easily overpower him. His position captures the country’s imagination. He is the representative of no constituency, but of the whole people. When he speaks in his true character, he speaks for no special interest. If he rightly interprets the national thought and boldly insists upon it, he is irresistible; and the country never feels the zest for action so much as when its President is of such insight and calibre.”

The BBI majority decision, however, looked beyond the constitutional silence in Articles 255 and 257 and, informed by the lived reality and history of the Kenyan people under the shackles of presidential imperialism, chose to bar the President from the popular initiative. Much as one would disagree with the conclusion of what the constitutional silence should have yielded in that case, one can easily see the constitutional realism that was at play and the weight with which it informed the majority decision. In other words, it is possible that without the historical context of Articles 255 to 257, Justice Njoki Ndungu’s SCJ dissenting opinion would have carried the day.

Just like BBI (specifically the popular initiative question vis-à-vis imperialism), the NGCDF debate sits at the intersection of constitutional fidelity and political demand. BBI had the unenviable task of deciding the limits of the most popularly elected leader in pushing for constitutional amendments through the popular initiative. The Court had an option to yield to political demand, but it chose to prioritise constitutional fidelity (in the manner it perceived best, whether we agree with it or not). That was important because it established a precedent and a judicial imprimatur that courts can draw institutional boundaries.

Now comes the NGCDF, and again the Supreme Court will have to strike a balance between constitutional fidelity and political demand regarding the first delegation of the people’s sovereignty, i.e., Parliament. On the one hand, it may be very difficult for the Supreme Court to wish away the political culture that equates representation with direct project implementation. For starters, striking down NGCDF without addressing (or at least acknowledging) the accountability and visibility gap it fills risks cosmetic redesigns and continued political resistance. On the other hand, upholding it in the same manner as the Court of Appeal’s NGCDF Act Judgment does, entrenches structural constitutional tensions. The Supreme Court is effectively embroiled in the classic “damned if you do, damned if you don’t” dilemma.

That is why, when the Supreme Court determines this dispute, it must be alive to the reality that what is at stake is: if the constituency is allowed to evolve into a quasi-service-delivery unit under the banner of decentralisation, we must be honest about what that means for devolution, the separation of powers and the identity of Parliament itself. One can only hope that the case at the Supreme Court will directly confront these competing interests between constitutional fidelity and political demand. But what makes the Supreme Court’s task truly unenviable is the fact that the dispute involving the constitutionality of NGCDF is separate from the question of whether NGCDF delivers projects. What the Supreme Court must decide is whether we, the People, are willing to reshape the constitutional order to satisfy the politics of visibility. How it ultimately decides this question is a matter of wait and see.

It Depends on What the Meaning of the Word “Its” is: A Critique of the Kenyan Court of Appeal’s CDF Decision

February 12, 2026February 12, 2026 Gautam Bhatia1 Comment

For the last decade, the Kenyan courts have been grappling with the constitutionality of “constituency development funds” [“CDF”]: that is, funds earmarked from the national government revenue for developmental work on a constituency basis. Judicial decisions have seen multiple repeals and amendments of the original CDF Act of 2013. These alterations to the CDF framework have changed the form, structure, and operation of the funds, while preserving their fundamental nature.

The latest salvo in this battle came from the Court of Appeal which, on 6th February 2026, upheld in significant part the NG-CDF Act of 2015, as amended in 2022 and 2023. In doing so, the Court of Appeal set aside the 2024 judgment of the High Court, which had struck down NG-CDF in its entirety (see here). On one limited issue, the Court of Appeal did agree with the High Court: that is, anchoring the term of the Fund Account Manager to the term of the Parliament was unconstitutional. However, the Court held that this provision could be severed from the rest of the NG-CDF Act, which would remain on the statute book.

Before critiquing the Court of Appeal’s decision, let me briefly recap the chronology of important events that brought us to this point.

The initial CDF Act, 2013, placed the constituency development funds under the effective control of constituency MPs. The 2013 Act was struck down by the High Court in 2015, primarily on two grounds: that it violated the constitutional provisions on devolution by setting up a “third tier” of government, and that it violated the separation of powers by turning legislators into constituency administrators.
In response to the judgment, parliament brought in the NG-CDF Act of 2015, which sought to address the High Court’s concerns by (a) creating an adequate separation between the MP and the Constituency Committees responsible for implementing the CDF at the constituency level; and (b) limiting the CDF to developmental projects within the domain of the national government.
Meanwhile the constitutional challenges to the now-defunct 2013 Act wound their way through the Court of Appeal and finally to the Supreme Court. The Supreme Court, in 2022, affirmed the High Court’s decision, and laid down some crucial constitutional principles: in particular, that the constituency could not be treated as a “service delivery unit,” and that the function of the legislature in a parliamentary structure was to engage in legislation, deliberation, and oversight – not administration (see here).
In 2022 and 2023, further amendments were made to the NG-CDF Act, to ensure greater insulation of the CDF from devolution and separation of powers challenges.
In 2024, the High Court – before which a prior challenge to the 2015 NG-CDF Act was pending – struck down the updated law as unconstitutional, holding that even the new(est) version of the Act continued to offend the principles of devolution and separation of powers, as outlined in the Supreme Court’s 2022 judgment.
On 6th February 2026, the Court of Appeal reversed the High Court’s judgment and upheld the NG-CDF 2015 (as amended in 2022 and 2023), subject to severing one of its provisions.

It should be clear from this account that the key question about the CDF has always been whether a “development” fund that takes the constituency as its unit, and remains connected in some way with the legislative arm of the State, is compliant with the principle of devolution and the separation of powers.

How did the Court of Appeal deal with these two issues? It held, first, that the NG-CDF was about decentralisation, and about ensuring access to the central government in all parts of Kenya. Thus, it was consistent with devolution. Secondly, it held that the under the present structure, the National Assembly was only exercising oversight functions over the CDF, and that this was consistent with the flexible vision of the separation of powers set out under the Kenyan Constitution.

Let us examine each of these justifications in turn.

On the devolution question, in paragraph 72, the Court of Appeal held that the Constitution required the central government to ensure reasonable access to its services in all parts of the republic, and the principle of subsidiary allowed this to be done through centralisation. This is entirely correct. However, with respect, it is also entirely besides the point. The question is not whether the central government can decentralise its functions, but whether it can do so through the constituency unit. The State was aware of this, because it specifically argued – as recorded by the Court in paragraph 60, that ministries may decentralise the functions of the government “to any level of its administrative units, including constituencies.”

But things are not quite as simple: with apologies to Bill Clinton, the correctness of this submission depends on what the meaning of the word “its” is. Can it be said that the constituency is an administrative unit of the central government (“its”?) The Court of Appeal appeared to say so, on two grounds. First, it held that the respondents’ argument that a constituency was not a unit of service delivery was based upon the basic structure of the Constitution, a concept that had been rejected by the Supreme Court in the BBI Judgment (paragraph 68). With respect, this confuses two separate ideas: the basic structure and the basic structure doctrine. What was (arguably) rejected by the Supreme Court in the BBI Judgment was the basic structure doctrine, i.e., which subjects constitutional amendments to judicial review on the touchstone of the Constitution’s basic structure. What was not rejected by the Supreme Court was the incontrovertible fact that the Constitution has a basic structure (one of whose features is representative democracy, which functions through constituency units). The present case did not involve a challenge to a constitutional amendment, and therefore, the Court of Appeal’s observations about the basic structure are inapposite.

Secondly, the Court of Appeal cited Article 89 of the Constitution to hold that “it is a unit firmly embedded within the national governance framework.” (paragraph 71) There are two problems with this observation. One, it is a mere assertion, not backed up by any reference to the constitutional text or structure. There is a vast gap between the fact that the constituency unit exists in the Constitution, and that it exists as an element of national governance. There are many things in the Constitution that are not part of “national governance.” And two, this observation is in the teeth of the Supreme Court’s explicit holding that the constituency is a unit of representation, and not of service delivery. There is a crucial, conceptual distinction between representation and governance: the first has to do with the composition and activities of the legislature, and the second with the role and functions of the executive. The Court of Appeal’s invocation of Article 89 simply elides this difference.

It should therefore be clear that, following the Supreme Court’s holding that the constituency is a unit of representation and not of service delivery, any legislative version of the CDF that retains the constituency as its basis has to be unconstitutional. It cannot be otherwise. The CDF is incontrovertibly about service delivery, and the constituency is not a service delivery unit: individually, these two propositions are unassailable, and together, they make the 2015/2022/2023 version of the NG-CDF as unconstitutional as the 2013 version, regardless of any other changes. The Court of Appeal tried to get around this by referring to the NG-CDF Act’s conceptualisation of the constituency, but with respect, this puts the cart before the horse: the nature of the constituency as a unit of representation flows from the constitutional scheme and design, and is upstream of legislation. Legislation, by stipulating that the constituency is a unit of service delivery, cannot thereby make it so.

Let us turn to the separation of powers. Even though the Court of Appeal substantively agreed with the High Court on the unconstitutionality of the term of the Fund Account Manager, it laid out a more deferential account of the separation of power, which we must scrutinise. The Court of Appeal held that separation of powers did not require “separation of arms and organs into impermeable silos but rather structured interaction anchored in accountability.” (para 95) This is undebatable – in fact, it is so undebatable that it is a truism that tells us nothing meaningful. Nobody has ever claimed that separation of powers requires “silos”: what is needed is an account of what separation of powers does require, and that – in turn – is grounded in questions of why a particular Constitution has the separation of powers. In the specific case of Kenya, as I have argued in Chapter Four of this book, the 2010 Constitution was framed as a response to a colonial and post-colonial history where representative bodies had been systematically degraded, and made subordinate to the executive. The separation of powers in Kenya is thus best understood as both transformative and restorative (per Dixon and Landau) – that is, restoring their core functions to the separate organs of state.

This perspective helps us understand the flaw in the Court of Appeal’s approach: while the Court of Appeal was correct to say that the separation of powers does not require State organs to function in “splendid isolation or tortured loneliness,” (para 97), there is a distinction between overlap that involves organs interacting in the process of checking or overseeing each other, and overlap that involves organs performing each other’s core functions. Indeed, this last bit was precisely what the Supreme Court, in its CDF judgment, held was the quintessence of the separation of powers doctrine, and this is something that the Court of Appeal did not engage with. Therefore, the National Assembly’s involvement with the CDF remains problematic in that the legislative organ is being asked to perform a core executive function. Crucially, it is important to recognise that the problem does not go away if the individual MP’s control over their constituency fund is taken away. Yes, this does address a particularly crude kind of incumbent bias and concentration of power, but the problem is also on the level of the legislative organ as a whole: that is, a legislative organ that takes, as its primary – or at least, equivalently important – task the management of CDF, is a legislature that risks diluting its own core functions of legislation, deliberation, and oversight. In other words, the problem with encroachment is not only that one State organ takes over the core functions of another, but that in doing so it becomes less able or willing to perform its own, very distinct core functions.

On the specific point, this distinction may not have made a difference to the outcome, as the Court of Appeal also found the Fund Account Manager provision to be unconstitutional. However, there is a great deal of daylight between the High Court and the Court of Appeal in how separation of powers is to be understood, which certainly will make a difference to the outcome of future cases. It is respectfully submitted that the High Court’s understanding is correct, both in terms of the Supreme Court’s 2022 judgment, and in terms of consistency with the Kenyan Constitution’s text, design, and historical influences in mandating the separation of powers.

The action now moves to the Supreme Court where, once again, the meaning of core ideas such as devolution and the separation of powers – and indeed, the true meaning of the Supreme Court’s own 2022 judgment – will be debated. It will be interesting to see how the Supreme Court resolves these questions, which go to the heart of Kenya’s transformative 2010 Constitution, and how it organises, shapes, and constrains public power.

Constitutional Law and Philosophy

Issues in Contemporary Constitutional Law, with a Special Focus on India and Kenya