The Right to Information Act was narrowed twice, and the two narrowings could hardly have arrived by more different roads. The first, the RTI Amendment Act 2019, was a short amending statute aimed openly at the men and women who run the Information Commissions: it took tenure and salary out of the parent Act and handed them to government rules. The second travelled in disguise. Section 44(3) of the Digital Personal Data Protection Act 2023, a sub-section inside a statute about data, substituted a single clause of the RTI Act and deleted, in one stroke, the public activity test, the privacy balancing test and the proviso that tied a citizen’s access to Parliament’s own. The first narrowing was debated for a week and denounced on the floor of both Houses. The second passed inside another bill, in a session consumed by a no-confidence motion, and was widely understood only after it was law. An amendment trail usually shows a statute drifting in one direction under steady pressure. This trail shows something more instructive: a statute can be narrowed from outside itself, by a provision that never has to survive the scrutiny a direct amendment attracts, and the outside route can cut deeper than the direct one.

The 2005 settlement: two choices that carried the structure

To see what the two amendments did, a reader first needs the design they worked on, because neither amendment touched the right itself. Section 3 of the Act of 2005 gives every citizen the right to information, and the machinery of request, time limit, first appeal, second appeal and penalty in Sections 6, 7, 19 and 20 was left intact by both narrowings. The full architecture of the statute, its definitions, its proactive disclosure duties and its exemption scheme, is set out in the complete guide to the Right to Information Act 2005, and this article takes that architecture as given. What concerns an amendment trail is the perimeter, and the perimeter of the 2005 Act rested on two deliberate design choices, each of which became a target.

The first choice was to insulate the adjudicators. The Information Commissions decide appeals against the very government departments whose officers staff the state, and the drafters of 2005 borrowed the standard constitutional technique for protecting such a body: fix the term in the statute, fix the salary in the statute, benchmark both against an office the executive cannot touch, and make removal difficult. Section 13 therefore gave the Chief Information Commissioner a term of five years from the date of entering office, barred reappointment, and set an outer ceiling at the age of sixty five. An Information Commissioner served five years or until sixty five, whichever came first, could not be reappointed to that office, and could be elevated to Chief subject to an aggregate ceiling of five years across both offices. Section 13(5) fixed salaries by equivalence: the Chief’s salary and allowances were the same as those of the Chief Election Commissioner, and an Information Commissioner’s the same as those of an Election Commissioner, with a set-off where the appointee already drew a pension from earlier government service. Section 16 made the corresponding arrangements for the State Commissions, with the State Chief Information Commissioner’s salary equated to an Election Commissioner’s and a State Information Commissioner’s to that of the Chief Secretary of the state government. A further safeguard ran through both sections: the conditions of service could not be varied to a commissioner’s disadvantage after appointment. Removal, under Section 14 at the Centre and Section 17 in the states, required proved misbehaviour or incapacity after an inquiry, or one of the stated grounds such as insolvency or conviction for an offence involving moral turpitude.

Why do fixed tenure and a fixed salary protect an adjudicator?

A fixed term and a salary written into the statute protect an adjudicator because the government of the day, the respondent in nearly every case the adjudicator decides, can neither shorten the term nor cut the pay in response to an unwelcome order, and cannot reward compliance either.

The protection lies in the location of the numbers: in an Act, they change only by another Act, debated in public; in rules, they change by notification. Independence in such a design is not a sentiment but a drafting address, and the whole argument over the 2019 amendment is an argument about that address.

The second design choice was to build a balance into the most sensitive exemption. Section 8(1) lists the categories of information exempt from disclosure, and clause (j) dealt with personal information. The clause as enacted did not exempt personal information as a class. It exempted information which relates to personal information the disclosure of which has no relationship to any public activity or interest, or which would cause unwarranted invasion of the privacy of the individual, and even then it allowed disclosure where the Central or State Public Information Officer or the appellate authority was satisfied that the larger public interest justified it. A proviso then added a floor beneath the whole scheme: information which cannot be denied to Parliament or a State Legislature shall not be denied to any person. Three distinct restraints were packed into that drafting. The exemption was triggered only where the information lacked a relationship to public activity or interest, or where disclosure would work an unwarranted, not merely an unwanted, invasion of privacy. It carried its own internal override, so that a sufficiently weighty public interest defeated the exemption inside the clause itself rather than depending on any provision elsewhere. And the proviso anchored the citizen’s entitlement to the legislature’s entitlement, on the reasoning that a democratic state cannot consistently claim, against its own citizens, a secrecy it does not claim against their representatives. Courts read the clause in that spirit. The Constitution Bench in Central Public Information Officer, Supreme Court of India v Subhash Chandra Agarwal, decided in 2019, treated the former clause as calling for a structured balancing between the right to information and the right to privacy rather than a categorical refusal, and the earlier decisions in Girish Ramchandra Deshpande in 2012 and Canara Bank v C. S. Shyam in 2017 resolved particular claims about service records and employee postings by applying its tests. Those authorities belong to the clause that existed before the substitution, and their fate is part of this trail.

The false starts: amendments proposed and abandoned

An amendment trail includes the amendments that failed, because they mark the points at which the political system tested the Act and recoiled, and the two narrowings that succeeded are better understood against the two that did not. The first attempt came within a year of commencement. In 2006, the government prepared a proposal to amend the Act so that file notings, the recorded opinions and advice of officers on a file, would be excluded from the definition of information except where they concerned development and social issues. File notings were the heart of the Act’s value to anyone tracing a decision, because they show who advised what and who overruled whom, and the proposal drew immediate and organised protest from the transparency movement that had campaigned for the Act itself. The proposal was not proceeded with. The episode established a pattern that held for over a decade: the Act’s text was politically expensive to touch directly, because its constituency was organised, articulate and watching.

The second false start came in 2013 and ran the same course. In June 2013, a full bench of the Central Information Commission held that six national political parties were public authorities under the Act, on the ground of their substantial public financing through land allotments, tax exemptions and other state facilities. The government of the day responded with the Right to Information (Amendment) Bill 2013, introduced in the Lok Sabha in August 2013, which would have amended the definition of public authority to exclude political parties and would have operated retrospectively to nullify the Commission’s order. The Bill was referred to the standing committee on personnel, the examination the 2019 Bill never received, and there it stayed; it lapsed with the dissolution of the Fifteenth Lok Sabha, and no successor government revived it in that form. The parties remained outside the Act in practice because they simply declined to comply with the Commission’s order and the Commission recorded its inability to enforce it, an enforcement failure the pillar article and the Commission’s profile both record, but the statutory text was not amended to bless the defiance. Read against this history, the interest of the 2019 amendment sharpens. The earlier attempts aimed at the Act’s substance, at what counts as information and who counts as a public authority, and both failed in the open. The 2019 amendment aimed at the service conditions of the adjudicators, ground on which the defending constituency was smaller and the classification argument available to the government was respectable, and it succeeded in a week. The 2023 substitution aimed at a single exemption clause from inside another statute, where the defending constituency was not assembled at all. The trail’s lesson about vulnerability is written across the sequence: the Act proved hardest to amend where its users could see the amendment coming.

The first narrowing: what the RTI Amendment Act 2019 actually changed

The Right to Information (Amendment) Act 2019 is Act 24 of 2019. It received the assent of the President on 1 August 2019 and was brought into force on 24 October 2019. For all the argument it generated, its text is short, and its method matters more than its length. The Act amended three provisions of the parent statute: Section 13, which governs the term and conditions of service of the Chief Information Commissioner and the Information Commissioners; Section 16, the corresponding provision for the State Chief Information Commissioner and the State Information Commissioners; and Section 27, the rule-making power of the Central Government, which was expanded so that the matters newly left to prescription could in fact be prescribed by rule.

In Section 13, the words that had fixed a term of five years were replaced, for the Chief and for the Information Commissioners alike, by the formula that a commissioner shall hold office for such term as may be prescribed by the Central Government. The sub-section that had fixed salaries by equivalence with the Election Commission was replaced by a provision that the salaries, allowances and other terms and conditions of service shall be such as may be prescribed by the Central Government. The safeguard against variation to a commissioner’s disadvantage after appointment was retained, and so were the bar on reappointment and the aggregate ceiling for a commissioner elevated to Chief. Critically, the outer age ceiling survived: the amendment did not remove the rule that a commissioner cannot hold office after attaining the age of sixty five, and any summary that reports the ceiling as abolished misstates the Act. Section 16 received the identical treatment for the state commissions, which carried a consequence the debate returned to repeatedly: the tenure and pay of a state commission’s members, a body constituted under a central Act but serving a state, would be prescribed not by the state government but by the Central Government.

What the amending Act conspicuously did not do was state the new term or the new salary. The numbers arrived three months later, in subordinate legislation. The Right to Information (Term of Office, Salaries, Allowances and Other Terms and Conditions of Service) Rules 2019, notified on 24 October 2019, the day the amendment commenced, prescribe a term of three years for the Chief Information Commissioner, the Information Commissioners, the State Chief Information Commissioner and the State Information Commissioners. They prescribe pay of Rs 2,50,000 fixed per month for the Chief Information Commissioner and Rs 2,25,000 fixed per month for the Information Commissioners and for the chiefs and members of the State Commissions, with dearness allowance and other facilities aligned to posts carrying the same pay and with a set-off of pension against pay for appointees who draw one. The term therefore fell from five years to three, and the salary ceased to be the salary of an Election Commissioner and became a figure named in a rule. Both halves of that sentence deserve emphasis, because commentary frequently collapses them. The reduction in the term was real, but it was effected by the rules, not by the amending Act, and the deeper change effected by the Act itself was jurisdictional rather than numerical: it moved the power to set the term and the salary from Parliament to the executive, permanently, for every future holder of these offices.

The amending Act, provision by provision

Because the amending Act is short, each of its moves can be isolated, and three of them reward separate attention. The first is the substitution in Section 13 of a prescribed term for a stated term. In the architecture of Indian public law, “as may be prescribed” is a term of art: it means the content of the law on that point will be supplied later, by rules, which are made by the executive, laid before Parliament, and almost never voted on. A statute that fixes five years and a statute that prescribes a term to be prescribed may produce identical outcomes in a given year, as they did not here, but they distribute power differently in every year, because the first can be changed only by the body that made it and the second by the body the statute regulates in its appellate capacity. The drafting choice also changed the burden of legislative inertia. Before 2019, a government that wanted shorter terms had to persuade Parliament to amend, and inertia protected the five-year term. After 2019, a government that wants any term at all need only notify it, and inertia protects the notification, since a rule stands unless Parliament takes the unusual step of annulling or modifying it on a motion.

The second move is the one least discussed and, in federal terms, the most striking. Section 27 of the parent Act is the Central Government’s rule-making power, and Section 28 is the corresponding power of the state governments for matters within their sphere, such as fees and the conditions of their own administrative machinery. The 2005 design left the service conditions of state commissioners in the Act itself, so in one sense no government, central or state, prescribed them. The 2019 amendment, by amending Section 27 to carry the prescription of terms for the State Chief Information Commissioner and the State Information Commissioners as well, placed the state commissions’ tenure within the Centre’s rule-making power. A state legislature cannot amend the central Act to recover the point, because the RTI Act is a central statute; a state government cannot prescribe a longer term for its own commissioners, because the power to prescribe is conferred on the Centre; and the state that constitutes the commission, appoints its members through the statutory committee and pays them from its own exchequer has no say in how long they serve or what they are paid. Whatever view a reader takes of the Election Commission parity question, this centralisation was not entailed by it, since parity could have been ended for state commissioners by benchmark figures stated in the Act just as it could for central ones. P. Chidambaram’s question in the Rajya Sabha, why the Centre should prescribe for a body the state constitutes, was therefore the federal form of the article’s central objection: the remedy travelled further than the rationale.

The third move is the survival, within the substituted text, of the protection against disadvantageous variation. Sections 13 and 16 continue to provide that a commissioner’s salaries, allowances and terms shall not be varied to the commissioner’s disadvantage after appointment. In the 2005 design that proviso was a backstop to a fixed statutory scheme. In the post-2019 design it is the only statutory protection a commissioner has, and its limits define the amendment’s practical meaning. It protects a sitting commissioner against a mid-term reduction aimed at or falling upon them. It does not protect the office against a general revision of terms between appointments, it does not prevent the prescribing authority from offering a shorter term to the next appointee, and it does not freeze the salary of a commissioner elevated from one office to another on terms the rules choose to set. A protection that operates only person by person and only during a term already granted is a different species of security from a term and salary stated in the Act, and the difference is not a nuance of drafting but the substance of what the amendment trail records.

A week in July: the passage of the 2019 Bill

The amending Bill was introduced in the Lok Sabha on 19 July 2019 by Dr Jitendra Singh, the Minister of State for Personnel, Public Grievances and Pensions, the ministry that administers the Act. The Lok Sabha passed it on 22 July 2019. The Rajya Sabha took it up on 25 July 2019, and the contest there was the real one, because the government did not command a majority in the Upper House and the opposition concentrated its effort on a procedural demand rather than on the merits alone: that the Bill be referred to a select committee of the House for examination before passage. The motion for reference was put to a vote and defeated, with 117 members voting against reference and 75 voting for it, a margin of 42 votes. The sitting was turbulent; opposition members protested during the voting itself, alleging that ruling party members had attempted to influence how some members voted, the Leader of the Opposition, Ghulam Nabi Azad, accused the government of converting Parliament into a department of the executive, and the Congress walked out before the Bill was put to its final vote. The Rajya Sabha passed the Bill the same day. Assent followed on 1 August 2019. From introduction to assent, the entire legislative journey occupied thirteen days, and the Bill spent seven of them in Parliament.

Two features of that record matter for the trail. The first is what did not happen. The Bill was not examined by the department-related standing committee on personnel, public grievances, law and justice, the committee that would ordinarily have taken evidence on a change to the service conditions of a statutory adjudicatory body, and the select committee route, which the opposition sought precisely because the standing committee route had not been used, was voted down. Reporting of the government’s own file notings, published subsequently, indicated that the presiding officers of both Houses had been asked to expedite the Bill’s passage and to relax the usual notice requirements for its introduction, which is why the Bill could be introduced and passed within a single week. The second feature is who objected, and in what capacity. Inside Parliament, P. Chidambaram pressed the federal objection: a state government constitutes the State Commission, so by what logic does the Centre prescribe the tenure and conditions of its members, and can the states not be trusted with appointments their own statute-level bodies require? Outside Parliament, former information commissioners, among them Wajahat Habibullah, the first Chief Information Commissioner, wrote to the Prime Minister urging that the Bill be withdrawn, arguing from their own experience of the office that security of tenure was a working condition of fearless adjudication rather than a perquisite. The Bill’s passage is therefore part of the article’s evidence, not merely its chronology: the scrutiny the measure received was a single week of floor debate in each House, with committee examination refused at both available points.

The government’s case, and the answer to it

The stance this series requires is that each side be stated at its strongest before either is judged, and the government’s case on the 2019 amendment is neither empty nor dishonest. Its core, as Dr Jitendra Singh put it in both Houses, is a classification argument. The Election Commission is a constitutional body, created by Article 324 of the Constitution, conducting elections to Parliament, the state legislatures and the offices of President and Vice President. The Information Commissions are statutory bodies, created by an Act of Parliament in 2005. To equate the service conditions of the head of a statutory body with those of a constitutional authority was, on this view, an anomaly embedded in the original drafting, a category error that the amendment corrected. The minister assured both Houses that the amendment carried no design to curtail the autonomy of the commissions, that it would streamline the working of the Act, and that the government would not exercise the prescribing power arbitrarily or vary terms capriciously; he resisted the charge that the power would be used to revise terms every couple of years as a speculation the text did not support. The broader defence, made in the government’s written statements around the Bill, was one of uniformity: service conditions across the many statutory commissions and tribunals in Indian law vary widely, and bringing the Information Commissions within a prescribable framework aligned them with the general run of statutory bodies rather than leaving them benchmarked to a constitutional office.

Stated at its strongest, that argument succeeds on its chosen ground. Parity with the Election Commission was an unusual drafting choice, and a legislature that wishes to grade its statutory bodies by function rather than by a borrowed constitutional benchmark is making a defensible policy decision. The verdict this article reaches is the sharper one the brief for this trail demands: the objection is not to the correction but to the remedy. If the anomaly was the benchmark, the remedy consistent with the stated purpose was to fix a different benchmark in the statute itself, a term of years and a salary stated in the Act and benchmarked to some other office or stated as figures, so that the numbers remained where the 2005 Parliament had put them, inside primary legislation. The remedy actually chosen achieved the stated correction and something further that the stated correction did not require: it relocated both numbers into rules made by the Central Government, which is the principal respondent before the commissions, the authority whose ministries, departments and public sector undertakings generate the appeals the commissions decide. A three-year term prescribed by rule can be shortened by rule, lengthened by rule, or left conspicuously unamended as a commissioner’s term runs, without any of those choices ever being debated on the floor of either House. The safeguard against mid-term variation to an incumbent’s disadvantage protects the individual sitting commissioner; it does nothing to protect the office, because the terms offered to the next appointee can be rewritten between appointments. That is why critics inside and outside Parliament, including former commissioners who had held the office, described the amendment as converting a statutory entitlement into an executive prescription. The classification argument answers the question why parity might end. It does not answer the question why the replacement had to be a delegation, and on the record of the debate, that second question was asked repeatedly and never met on its own terms.

The uniformity argument meets its counterexample

The government’s uniformity defence, that the Information Commissions were being aligned with the general run of statutory bodies, had a specific weakness that the Bill’s opponents identified within days, and it came from the government’s own recent practice. Deepak Sandhu, a former Chief Information Commissioner and the first woman to hold that office, pointed out in opposing the Bill that in 2017 the same government had carried out a rationalisation of the salaries of some nineteen statutory bodies and had placed the heads and members of seventeen of them at equivalence with the Chief Election Commissioner and the Election Commissioners respectively. If parity with the Election Commission was an anomaly to be corrected, it was an anomaly the government had itself conferred or confirmed across the statutory landscape two years earlier, and the question her intervention posed, why the Information Commission was identified and removed from that entire lot, was not answered in the debate. She added a procedural objection of the same character: that the Bill had been brought without the pre-legislative consultation that the government’s own consultation policy, adopted in 2014, contemplated for measures of this kind, so that the persons who had held the offices being redesigned learned of the redesign from the Bill itself. Other former commissioners spoke in the same week and to the same effect: M. M. Ansari argued the amendment’s purpose and tendency was to dilute the Commission, M. Sridhar Acharyulu, who had sat as an Information Commissioner, urged that the Act not be touched and the Bill be withdrawn, and Annapurna Dixit joined the demand for withdrawal.

These interventions matter to the trail for a reason beyond their eminence. The debate about the 2019 amendment is usually reported as a contest between a government and an opposition, which makes it legible as ordinary politics and lets a reader discount both sides as interested. The former commissioners were not an opposition. They were the office’s previous occupants, drawn from the civil services and the professions the Act names, several of them appointees of governments of different complexions, and their objection was institutional rather than partisan: that a body which adjudicates against the executive cannot have its members’ terms set by the executive without the adjudication itself being placed under a shadow, whatever the intentions of the government of the day. Set beside Sandhu’s counterexample on the 2017 rationalisation, the uniformity argument loses its factual premise as well as its institutional one, and the verdict recorded in the previous section, that the correction was defensible and the remedy was not, stands on the government’s own record rather than on the critics’ rhetoric. It is also the point at which the article’s method should be visible to the reader: where a government’s rationale can be tested against a document, a date or a prior act of the same government, the test is worth more than another exchange of characterisations, and on the two occasions this trail has run that test, the classification argument in 2019 and the harmonisation argument in 2023, the rationale has survived in part and the remedy has not survived at all.

What the 2019 amendment left standing

An amendment trail that records only movement misleads, so the continuities deserve their own account. The removal provisions survived untouched: a commissioner can still be removed only for proved misbehaviour or incapacity after an inquiry by the Supreme Court on a reference, or on the stated grounds in the Act, and the executive acquired no power of removal at pleasure. The bar on reappointment survived, which matters more than it appears to, because a commissioner who cannot be reappointed has no second term to earn by pleasing the appointing authority. The aggregate ceiling on elevation from Information Commissioner to Chief survived. The age ceiling of sixty five survived. The powers of the commissions under Sections 18, 19 and 20, to inquire with the powers of a civil court, to direct disclosure, to order compensation in the cases the Act specifies and to impose a penalty of two hundred and fifty rupees a day, capped at twenty five thousand rupees, on an officer who has refused or delayed without reasonable cause, all survived word for word. The institution whose composition, appointment process, powers and structural constraints are examined in the profile of the Central Information Commission remained, on paper, the same adjudicatory body the day after the amendment as the day before. What changed was narrower and, for that reason, easier to underestimate: the terms on which its members hold office ceased to be a promise made by Parliament in the statute and became a prescription made by the executive in rules, beginning with a term shortened from five years to three. When the commissioners’ tenure is discussed, the institutional ownership of that question sits with the Commission’s own profile; what belongs to this trail is the mechanism, a statutory term converted into a delegated one, and the mechanism is the amendment’s lasting contribution to the Act’s history.

The figures themselves: what the prescribed pay did and did not cut

The salary half of the 2019 Rules repays the same provision-level attention as the tenure half, because the cash figures complicate the usual account in a way that strengthens rather than weakens the article’s verdict. Under the former Section 13(5), the Chief Information Commissioner’s salary was that of the Chief Election Commissioner, whose salary is by law equal to that of a judge of the Supreme Court, a figure that stands at Rs 2,50,000 a month under the salaries legislation as revised in 2018. The 2019 Rules prescribe the Chief Information Commissioner a fixed pay of Rs 2,50,000 a month. In cash terms, therefore, the Chief’s opening figure under the rules is the same figure parity would have produced, and a critic who says the amendment cut the Chief’s salary is wrong on the arithmetic. The Information Commissioners’ position is different in degree: an Election Commissioner’s salary, likewise pegged to a Supreme Court judge’s, is Rs 2,50,000 a month, and the Rules prescribe Information Commissioners, and the chiefs and members of the State Commissions, Rs 2,25,000 a month, which is Rs 25,000 a month below the figure the former parity, or in the State Chief’s case the former Election Commissioner equivalence, would have carried.

What the figures show, once they are stated exactly, is that the amendment’s salary change was never principally about the opening number. It was about the mechanism that produces the number over time. A salary fixed by equivalence is a salary that revises itself: when Parliament revises the salaries of Supreme Court judges, as it did in 2018, the Election Commissioners’ salaries move with them, and under the former law the Information Commissioners’ salaries would have moved in the same motion, without any decision by the government of the day and without any possibility of the government of the day deciding otherwise. A salary fixed by rule at Rs 2,50,000 or Rs 2,25,000 moves only when the prescribing authority chooses to move it, and stays frozen, in money that inflation steadily thins, for exactly as long as the authority is content to leave it. The equivalence was also a statement of rank in the order of precedence of the Republic’s institutions, which is not a vanity: the rank of the adjudicator is part of what a department’s senior officers register when an order arrives from the Commission, and the 2005 Parliament had deliberately registered the commissions at the Election Commission’s rank. The rules replaced a self-revising equivalence and a rank with two frozen figures and no rank at all. Stated with the arithmetic on the table, the salary story is the tenure story in miniature, and it is why this article has insisted throughout on precision about what changed: not, in the main, the first number a commissioner was paid, but the identity of the authority that would decide every number after the first.

Who the three-year term actually bound: the transition question

A trail should record the transition an amendment creates, because amendments operate on offices already filled and rights already accrued, and the 2019 amendment’s transition was governed by the very proviso the debate overlooked. When the amending Act commenced on 24 October 2019 and the Rules prescribed a term of three years on the same day, the commissioners then in office at the Centre and in the states had been appointed under the unamended Act, on terms the statute had fixed at five years and salaries the statute had benchmarked. Could the new prescription reach back and shorten a sitting commissioner’s term to three years, or recompute a sitting commissioner’s salary? The retained safeguard answers the question in terms: the salaries, allowances and other terms and conditions of service of a commissioner shall not be varied to the commissioner’s disadvantage after appointment. A term already granted under the Act is a term of service within that protection, and a rule that cut it would vary the commissioner’s conditions to their disadvantage in the most literal way available. The three-year term therefore governed the future, appointments made after the commencement, while the incumbents of 2019 served out the terms the old law had promised them. The same logic protected sitting commissioners’ salaries against reduction, though not, it should be noted, against the quieter effect that the benchmark for any future elevation or reappointment conversation had changed beneath them.

The transition matters beyond fairness to a handful of officeholders, because it reveals the amendment’s tempo. A government in a hurry to change an institution’s behaviour through its personnel must ordinarily wait for the institution to turn over, and a five-year protected incumbency is a long wait. The 2019 design accepted that wait for the sitting generation and pre-committed every succeeding one, which is the signature of a structural rather than a retaliatory amendment: its object was not the commissioners who had given unwelcome orders, whose terms it could not touch, but the office itself, whose terms it permanently relocated. Critics who described the amendment as an attack on particular commissioners misdescribed it, and defenders who pointed to the incumbents’ undisturbed terms as proof of benign intent equally missed the design. The trail’s judgment stands between the two descriptions: the amendment was patient, general and permanent, and it is precisely those qualities that make its relocation of power the significant event, since a temporary or personal measure would have expired with its occasion and this one compounded.

Three years in practice: what a prescribed term does to an institution

The consequences of the 2019 amendment play out over years rather than at commencement, and they follow from the interaction of a shorter term with the appointment process the Act elsewhere prescribes. A commission’s working strength depends on appointments being made before vacancies arise, because a second appeal is heard by a bench of the commissioners in office and no provision of the Act lets a retired commissioner finish a part-heard docket. Under a five-year term, an appointment cycle that runs late costs the institution a fraction of a long tenure. Under a three-year term, the same delay consumes a larger fraction of each commissioner’s tenure, the turnover of benches is faster, part-heard matters change hands more often, and a government that simply does nothing, that leaves posts unfilled while a term runs down, achieves a smaller commission without any act that a court or a legislature can easily fasten on. The Commission’s own record of vacancy and pendency, documented with its sources in the profile of that institution, shows appointment delay as the standing condition of the system rather than an aberration, so the interaction is not hypothetical: shortening the term in a system that appoints late multiplies the effect of the lateness.

The second practical consequence concerns the kind of person the office can attract and the kind of independence it can sustain. A five-year term at a constitutional benchmark salary, with reappointment barred, made the office a plausible final appointment for a senior judge, a retired civil servant of standing or a person of eminence in the fields Section 12 names, who could take it up without calculating a future. A three-year term at a prescribed salary, renewable neither in law nor, given the reappointment bar, in expectation, compresses the office’s horizon: a commissioner who joins a docket measured in years of pending appeals knows the term will end before the docket does, and the institution’s memory, its accumulated sense of how departments resist and which arguments recur, resides in a registry and a secretariat rather than in the bench. None of this proves that any individual commissioner’s orders softened after 2019, and this article makes no such claim, because the evidence for it does not exist in a form that would support the assertion. The structural point is the one the critics made in advance and the design literature supports generally: tenure rules are written for the office across all its future holders, not for the holder of the moment, and a rule that makes the office shorter, less predictable and dependent on the respondent’s prescription is a rule that has changed what the office is, whether or not any single decision can be traced to it.

The contrast of the same season: when Parliament legislates the number itself

The 2019 RTI amendment was not the only occasion on which the Parliament of that period revisited the tenure of a watchdog body, and the comparison Parliament itself supplied is the cleanest evidence that the delegation, rather than the shortening, was a choice. In the same year, the Protection of Human Rights (Amendment) Act 2019 reduced the term of the Chairperson and members of the National Human Rights Commission from five years to three years, the identical numerical journey the RTI Rules would take the information commissioners on. But there, Parliament wrote the three years into the parent Act itself, where any future Parliament that disagreed would have to amend the statute in the open, and it coupled the reduction with changes to eligibility that were likewise enacted in the text. The human rights commission’s members therefore hold a shorter term than their predecessors and hold it on the same legal footing as before: a statutory entitlement, defeasible only by statute. The information commissioners hold a shorter term on a different footing altogether, an entitlement that exists at the level of rules and can be revised at that level. Two commissions, two reductions of two years each, two opposite answers to the question of where a term should live. A reader who wants to know what the RTI amendment was really about is better served by that comparison than by any speech in the debate, because it isolates the variable: the number was not the controversy, since Parliament cheerfully enacted the same number elsewhere in the same season; the location was the controversy, and the location was chosen.

The wider statute book tells the same story from the other side. The tenure of the Central Vigilance Commissioner is stated in the Central Vigilance Commission Act, the tenure of the members of the Competition Commission in the Competition Act, the tenure of the Comptroller and Auditor General in the Constitution itself, and in each case the number’s home is the instrument that created the office, so that changing it requires the same solemnity that created it. The 2005 Parliament followed that convention for the Information Commissions and improved on it with the Election Commission benchmark. The 2019 Parliament departed from the convention for this one set of offices while continuing to observe it elsewhere, and departed, moreover, in favour of the department that administers the Act and answers the appeals. Departures from a drafting convention are sometimes reforms and sometimes tells; what makes this one a tell is that the convention’s purpose is precisely to keep a respondent’s hands off an adjudicator’s terms, and the departure placed the terms in the respondent’s hands.

The clause the courts built, before it was replaced

Because the substituted clause will be read for years against the memory of the former one, the judicial construction of the former clause deserves a fuller account than a passing reference, both for what it shows about the old design and because the pending challenge asks the Supreme Court to say how much of that construction survives. The former clause’s working life began in the commissions, where officers applied its three tests to the unglamorous run of requests, and its leading judicial statements came in three decisions that between them mapped the clause’s possibilities.

Girish Ramchandra Deshpande v Central Information Commissioner, decided in 2012, was the restrictive pole. An applicant had sought the service record of a public servant, the details of disciplinary proceedings, and information about the officer’s assets, gifts and income tax returns. The Supreme Court held that such material was personal information within clause (j), that performance in public service, disciplinary history and assets were matters personal to the officer unless a larger public interest in disclosure was shown, and that income tax returns in particular were not to be disclosed on a stranger’s request. The decision was criticised at the time, and since, for the breadth of its equation between an officer’s public functions and his personal sphere, but its structure is the point for this trail: even at its most restrictive, the former clause required the court to pass through the tests, to ask after relationship to public activity and to leave the larger public interest inquiry standing at the end. An applicant who lost in Deshpande lost after the balancing had been performed and recorded. The substituted clause dispenses with the performance.

Canara Bank v C. S. Shyam, decided in 2017, applied the same structure to an employee’s postings, transfers and relieving orders in a public sector bank, and held them to be personal information under the former clause. Read together, the two decisions show the former clause at work in the employment setting where applicants most often met it, and they cut in favour of the officer as often as against him. What they never did was treat the presence of personal information as the end of the case, because the clause they were construing did not permit that treatment.

Central Public Information Officer, Supreme Court of India v Subhash Chandra Agarwal, decided by a Constitution Bench in November 2019, is the construction that matters most, because it was delivered after Puttaswamy had constitutionalised privacy and it set the two rights in a deliberate relation. The immediate question was whether the office of the Chief Justice of India is a public authority under the Act, which the Court answered affirmatively, but the Bench’s treatment of clause (j) ranged wider: it held that the right to information and the right to privacy are not to be played off as a zero-sum contest, that disclosure questions under the clause call for a structured inquiry into the nature of the information, its relationship to public activity and the proportionality of the intrusion, and that judicial independence, the value pressed into service against disclosure in that case, does not require opacity about the institution’s administrative functioning. That is the high-water mark of the balancing design, reached four years before the design was deleted. The petitioners in the pending challenge lean on Agarwal for the proposition that the Constitution Bench had already supplied the method by which privacy and transparency are reconciled, so that a substitution which abolishes the reconciliation rather than applying it cannot claim Puttaswamy as its warrant. The Union’s answer, that a legislature may choose a brighter line than a balancing test, is the contest the larger bench will resolve, and the trail records the sequence without anticipating it: a clause built for balancing, construed as a balance by a Constitution Bench, replaced by a threshold while the balance’s authority was still fresh.

The second narrowing: a clause inside a data protection statute

The second narrowing was aimed not at the adjudicators but at a single exemption, and it reached the statute book as Section 44 of the Digital Personal Data Protection Act 2023, the DPDP Act, which is Act 22 of 2023. Section 44 is the DPDP Act’s amendment section, the place where that statute tidies its boundaries with earlier legislation, and sub-section (3) provides that in the Right to Information Act, in Section 8, in sub-section (1), for clause (j), a new clause shall be substituted. The substituted clause reads, in full, “information which relates to personal information”. The parent statute’s own provisions, its definitions, its Data Protection Board and its penalty regime, belong to the article on the Digital Personal Data Protection Act 2023, and this article does not re-cover them; its subject is the single sub-section that crossed into another statute and rewrote it.

Set the two texts side by side and the scale of the change becomes visible. The former clause exempted personal information whose disclosure had no relationship to any public activity or interest, or which would cause an unwarranted invasion of the privacy of the individual, subject to the satisfaction of the Public Information Officer or the appellate authority that the larger public interest justified disclosure, and subject to the proviso tying the citizen’s right to Parliament’s. The substituted clause exempts information which relates to personal information. Every qualifier is gone. A relational threshold, the information must relate to personal information, is the whole of the test. There is no inquiry into public activity or interest, because the words that required it were deleted. There is no inquiry into whether an invasion of privacy is warranted or unwarranted, because that distinction was deleted with them. There is no override inside the clause, because the words conferring it were deleted. And there is no proviso, because the proviso went with the rest of the sentence in which it stood. Nor does the RTI Act supply a definition that confines the new clause’s reach: the Act’s definition section defines information, record and public authority at length, but “personal information” is not a defined term in the parent Act, so the breadth of the substituted clause will be worked out in the orders of commissions and the judgments of courts applying an undefined phrase to everything from a beneficiary list to a muster roll to the file notings on a public contract.

Why does it matter that the RTI Act does not define personal information?

Because the substituted clause exempts whatever relates to personal information, and the parent Act contains no definition of that phrase, its reach cannot be read off the statute: it will be fixed by commissions and courts case by case, from a beneficiary list to muster rolls to file notings on public contracts, with no textual boundary to consult in advance.

The substitution did not take effect when the DPDP Act was enacted. Like much of that Act, Section 44(3) awaited commencement by notification, and it was brought into force on 13 November 2025 by notification G.S.R. 843(E), the same notification exercise that set up the Data Protection Board and brought the Digital Personal Data Protection Rules 2025 into operation. Between August 2023 and November 2025, therefore, the substituted clause existed as enacted law awaiting commencement, the former clause continued to govern applications, and a reader who quoted the former clause as the law during that interval was right on the text and wrong only about its future. From 13 November 2025, the short clause governs, and every reproduction of the long clause as the operative text, of which there are many in circulation, states a law that has been replaced.

A direct amendment is made by a bill whose title, objects and debate are all about the statute being amended, so the affected community, the responsible committee and the press know which law is on the table. A collateral amendment travels inside a bill about something else, and is examined, if at all, as part of the host’s story.

The amended statute’s constituency may learn of the change only after passage, and the debate, the committee and the headlines all belong to the host statute rather than to the law being changed. The legal effect of the two routes is identical, which is exactly why the difference in scrutiny matters: the same words, moved from a direct bill into a host bill, face a fraction of the examination.

August 2023: passage under the shadow of a no-confidence motion

The DPDP Bill was introduced in the Lok Sabha on 3 August 2023 by Ashwini Vaishnaw, the Minister for Electronics and Information Technology. The Lok Sabha passed it on 7 August 2023 and the Rajya Sabha on 9 August 2023, and it received the President’s assent on 11 August 2023. The monsoon session in which this happened was dominated by the opposition’s no-confidence motion against the government over the situation in Manipur, which was debated in the Lok Sabha across the very days, 8 to 10 August 2023, between the data protection Bill’s passage in the Lower House and its passage in the Upper. Opposition members demanded that the Bill be referred to a standing committee and protested that a measure of this reach was being carried by voice vote amid disorder, with many members’ attention and the country’s news agenda fixed on the confidence debate. The demand for committee reference was not accepted, and the Bill was not examined by a committee of either House.

That record invites a comparison the trail cannot avoid, because the subject of data protection had not lacked for scrutiny in its earlier avatars. A personal data protection bill introduced in 2019 had been sent to a joint parliamentary committee, which examined it for two years and reported with extensive recommendations, before that bill was withdrawn in 2022 and a fresh draft published for consultation. The enacted 2023 text was a new bill, not the committee’s bill, and it carried in its amendment section a change to the RTI Act that no committee had examined in any avatar: the joint committee’s work had addressed the relationship between data protection and transparency as a policy question, but Section 44(3)’s substitution of Section 8(1)(j) was a feature of the 2023 text, and it passed both Houses within six days of the Bill’s introduction. Gaurav Gogoi, speaking for the opposition parties that later organised against the provision, made exactly this point when the repeal campaign was launched: that the Bill had been passed at a moment when the nation’s attention was fixed on the no-confidence motion concerning Manipur, and that a change of this consequence deserved the deliberation it never received. Whether or not a reader accepts the opposition’s rhetoric about the manner of passage, the chronology is a matter of record, and the chronology is the scrutiny row of this amendment’s history: introduction on 3 August, passage in the Lok Sabha on 7 August, passage in the Rajya Sabha on 9 August, assent on 11 August.

Commencement day: the provision that took effect at once

The manner of the substitution’s commencement adds a final entry to its record, because the DPDP Act was designed for staged commencement and the stages were not chosen neutrally. Most of that Act’s operative obligations on data fiduciaries were deferred, to be phased in as the Data Protection Board was constituted and the Rules settled, with the principal compliance obligations applying only from May 2027 on the government’s own notified schedule. Section 44(3) was not deferred. When the commencement notification G.S.R. 843(E) issued on 13 November 2025, the substitution of Section 8(1)(j) was brought into force with immediate effect, in the same exercise that established the Board and brought the Digital Personal Data Protection Rules 2025 into operation. The one provision of the data protection statute that amended another statute was thus among the first provisions of the statute to operate, years ahead of the compliance regime that was the statute’s stated reason for existing, and it began to govern RTI applications, including the stock of requests and appeals built up under the former clause’s tests, while the litigation against it stood referred to a larger bench and the repeal campaign stood unanswered.

Nor was the substitution wholly unnoticed between enactment and commencement, and the record should not pretend otherwise. Opposition members who spoke against the DPDP Bill in the Lok Sabha in August 2023 named the RTI amendment among their objections, and the Editors Guild of India and the Press Club of India placed their concerns about the provision’s effect on journalistic work on record in the period that followed, since reporting that depends on records about named officials is the first casualty of a clause that registers only that information is personal. What the interval between August 2023 and November 2025 lacked was not objection but a forum: the provision was already law, the repeal demand required a government majority to move against itself, and the commencement decision, the last discretionary point at which the substitution could have been deferred pending the court’s answer, was an executive act taken by notification. A trail that follows a provision from a bill’s amendment section through a two-year dormancy to a commencement notification makes the general point one final time. The consequential moments in a collateral amendment’s life are all low-visibility moments, a clause in a schedule, a line in a commencement notification, and at none of them does the legislative process, as it actually operated here, require anyone to stand up and defend the change to the statute that is being changed.

One transitional question the commencement left open deserves recording, because thousands of live matters turn on it and the substituted text does not answer it on its face. On 13 November 2025, a large stock of applications, first appeals and second appeals stood pending at every stage of the Act’s machinery, each of them framed, argued and in many cases part-decided under the former clause’s tests. Whether a request filed while the former clause governed falls to be decided under that clause, on the principle that the applicant’s entitlement crystallised when the request was made, or under the substituted clause, on the principle that an authority applies the law in force when it decides, is a question on which the Act, the DPDP Act and the commencement notification are all silent. The general law of repeal and substitution supplies arguments both ways, commissions will have to answer the question case by case until a court settles it, and the applicants caught in the interval will find their outcomes depending on the accident of which side of a notification date their hearing fell. A legislature that had examined the substitution in committee would have had the transitional clause drafted for it as a matter of routine; that no transitional provision exists is the smallest and most telling entry in the scrutiny row.

The two narrowings compared

Aspect	The 2019 amendment	The 2023 substitution
Instrument	The Right to Information (Amendment) Act 2019, Act 24 of 2019, a bill whose sole subject was the RTI Act	Section 44(3) of the Digital Personal Data Protection Act 2023, Act 22 of 2023, one sub-section inside a statute about data protection
Provision touched	Sections 13, 16 and 27 of the RTI Act: tenure, salaries and the rule-making power	Section 8(1)(j) of the RTI Act: the personal information exemption
What it changed	Removed the statutory five-year term and Election Commission salary parity, and left both to be prescribed by Central Government rules, which then set a three-year term and fixed monthly pay	Replaced a qualified, balanced exemption with the single phrase “information which relates to personal information”, deleting the public activity test, the unwarranted invasion test, the internal public interest override and the Parliament proviso
Parliamentary time	Introduced 19 July 2019, passed by the Lok Sabha on 22 July and the Rajya Sabha on 25 July, assented to on 1 August 2019	Host bill introduced 3 August 2023, passed by the Lok Sabha on 7 August and the Rajya Sabha on 9 August, assented to on 11 August 2023
Committee scrutiny	No standing committee examination; a select committee was demanded in the Rajya Sabha and the motion for reference was defeated by 117 votes to 75	No committee examination of the enacted text; opposition demands for reference were not accepted, and the RTI change had not appeared in any earlier committee-examined draft
Who noticed at the time	The opposition in both Houses, former information commissioners who wrote to the Prime Minister, and sustained press coverage of a contested vote	Few outside the data protection debate; the RTI consequence was identified by transparency advocates during passage and reached wide public attention through the repeal campaign that followed
Practical consequence	Commissioners hold office on terms the executive prescribes and can re-prescribe, beginning with a term of three years instead of five	A public authority can refuse any information that relates to personal information without applying the balancing tests the former clause required, leaving Section 8(2) as the surviving route for public interest disclosure
Commencement	In force 24 October 2019, with the 2019 Rules notified the same day	Enacted in August 2023 and brought into force on 13 November 2025 by notification G.S.R. 843(E)

Read the table by its scrutiny row and the argument of this article states itself. The change that was debated for a week, voted on division and campaigned against in real time altered the conditions of the adjudicators’ offices. The change that crossed both Houses inside another bill, unexamined by any committee in its enacted form, removed the balancing machinery from the exemption that individual applicants meet most often. The row order in the table is deliberate: what a provision changes is a question of law, but how much examination it received is a question about the legislative process, and on this trail the second question predicts the first one’s consequences better than any commentary written at the time.

The collateral amendment as a general technique

The 2023 substitution deserves to be named as an instance of a general drafting technique, because a reader who can name it can detect it, and detection is the only reliable defence the legislative process has. A collateral amendment is a provision in a statute that amends a different statute, on a subject outside the host statute’s own field, typically placed in the host’s closing sections among its repeals and amendments, and drafted in the passive formulae of substitution and omission that draw no attention to themselves. The technique is not inherently illegitimate. Statute law requires constant consequential amendment, definitions must be aligned when a code replaces another, and a Parliament that forbade cross-amendment would legislate in a tangle. The vice the technique makes possible is narrower: a substantive change to a watched statute, of a kind that would attract a committee, a constituency and a controversy if proposed directly, can be carried in a host bill whose scrutiny is organised around a different subject, whose committee, if the bill goes to one at all, has the wrong expertise and the wrong witnesses for the collateral question, and whose press coverage is written by correspondents covering the host subject.

Indian legislative practice supplies the comparison that proves the point is general and not an artefact of this one controversy. The Finance Act 2017, a money bill, carried provisions that amended a series of other statutes to restructure the tribunals those statutes created, and empowered the Central Government to prescribe the qualifications, tenure and service conditions of tribunal members by rules, the same relocation of adjudicators’ conditions from statute to rule that the RTI amendment performed two years later. When the rules framed under that power were challenged, the Supreme Court in Roger Mathew v South Indian Bank, decided in 2019, struck the rules down, held that the conditions of service of adjudicatory bodies could not be left to the unguided prescription of the executive whose departments appear before them, and referred the larger questions about the money bill route to a larger bench. The parallel is instructive in both directions. It shows that the relocation technique was a settled instrument of drafting by the time the RTI amendment used it, applied across the tribunal system through a host bill chosen partly because a money bill cannot be amended or rejected by the Rajya Sabha. And it shows the one systematic check the system possesses operating after the fact: a court examining whether the delegation itself is constitutionally tolerable. The RTI’s 2019 amendment used the direct route and so faced its Parliament; the tribunal changes used the collateral route and faced, in the end, a Constitution Bench. The 2023 substitution of Section 8(1)(j) is the collateral technique applied to a disclosure statute, and the challenge pending against it asks, among its other questions, the question Roger Mathew asked about a neighbouring field: how much of an adjudicative or rights-bearing scheme may be rewritten by prescription and indirection before the rewriting itself becomes the constitutional problem.

For the legislator and the staffer, the detection discipline follows from the anatomy. Read the closing sections of every bill, the sections headed amendments, repeals and savings, with the same attention given to its opening sections, because consequential amendment clauses are printed there and a substitution formula names its target statute and section in terms that reward a reader who checks what the target text says before and after. Compare the host bill’s statement of objects and reasons against its amendment schedule, because a genuine consequential amendment is always explicable from the host’s own purposes, while a collateral one requires a justification the statement usually does not supply. And treat any substitution that deletes provisos and overrides, rather than adjusting figures or names, as substantive by definition, because machinery words are where balances live. These are not counsels of suspicion; they are the reading habits of any legislature that examines bills in committee, and they are stated here because on both narrowings in this trail, the habits were not applied in time.

Three deletions in one sentence

Because the substitution is usually described in a single phrase, a clause was simplified, its three deletions are worth taking one at a time, since each removed a different protection and each will be missed in a different class of case.

The first deletion is the public activity qualifier. Under the former clause, information about a person acting in a public capacity stood on a different footing from information about the same person in a private capacity, because the exemption bit only where disclosure had no relationship to any public activity or interest. The salary drawn from public funds, the assets declared by a public servant, the beneficiaries of a public scheme, the marks awarded in a public examination, the contractor paid from a public tender: all of these relate to identifiable individuals, and all of them could be, and routinely were, disclosed because their relationship to public activity took them outside the exemption or let the balancing run in favour of disclosure. Under the substituted clause, the relational question is the only question, and it runs the wrong way for the applicant: the more plainly information relates to an identifiable person, the more squarely it falls within the clause’s words, whatever its relationship to public activity.

The second deletion is the unwarranted invasion test, which was the former clause’s privacy protection in its calibrated form. The clause never protected privacy in absolute terms. It protected it against unwarranted invasion, which required the deciding officer to ask whether the disclosure’s intrusion into the individual’s private sphere was justified by the public character of the information, a proportionality inquiry of exactly the kind the Supreme Court’s privacy jurisprudence in Justice K. S. Puttaswamy v Union of India, decided by a nine-judge bench in 2017, would lead a reader to expect. Deleting the word “unwarranted” along with the rest of the phrase did not strengthen privacy protection in any principled sense; it replaced a test that weighed privacy against public interest with a threshold that registers only that personal information is present. A privacy interest that would have survived the old test, a purely private detail with no public dimension, is protected under the new clause. A privacy interest that the old test would have overridden, the private detail that is also the evidence of a public wrong, is protected under the new clause in exactly the same way, and that identity of treatment is the criticism in its precise form.

The third deletion is the proviso, and it is the least understood. The proviso provided that information which cannot be denied to Parliament or a State Legislature shall not be denied to any person. Its function was constitutional as much as practical. Ministers answer questions on the floor of the House; committees of Parliament call for records; a legislature’s entitlement to information from the executive is a foundation of responsible government. The proviso wrote the citizen into that entitlement, so that an authority refusing a request had to confront the question whether it would also refuse the same record to a legislature, and if it would not, it could not refuse the citizen either. Commentators who had never examined the clause rarely knew the proviso existed, which is why its removal passed with so little remark; but for applicants it operated as a floor argument of last resort in cases where the balancing had gone against them. Its deletion severs that link, and with it goes a distinctive feature of the 2005 design: the equation of the citizen’s statutory right with the representative’s constitutional one.

Why did the deleted proviso link the citizen to Parliament?

The proviso linked the citizen to Parliament by making the legislature’s entitlement to information the floor of the citizen’s entitlement: whatever the executive could not withhold from Parliament or a State Legislature, it could not withhold from any person either.

Its drafting rested on the theory of responsible government, that the executive answers to the legislature and, through it, to the citizen, so a secrecy good against the citizen but bad against the House was a constitutional anomaly the 2005 Parliament chose to forbid rather than merely discourage.

The privacy case, taken seriously

The substitution has a genuine rationale, and an account that treats it as pretext fails its own test. A data protection statute must say something about the processing of personal data by public authorities, because the state is the largest data fiduciary in the country, and the right to privacy recognised in Puttaswamy is a fundamental right that binds the state in its handling of citizens’ information. The two statutes genuinely intersected, and some reconciliation between a disclosure law written in 2005 and a privacy right articulated in 2017 was unavoidable work for a Parliament enacting a data protection code. The government’s stated position runs along these lines: that the amendment harmonises the RTI framework with the constitutional status of privacy, and that information which other laws require to be disclosed will continue to be disclosed, because the DPDP Act preserves processing and disclosure carried out under other legal obligations. The first half of that position is a real argument. The former clause’s balancing was administered by thousands of Public Information Officers of uneven training, its outcomes were inconsistent between authorities and sometimes within them, and a legislature could reasonably conclude that the protection of personal data by public bodies needed a clearer rule than a case-by-case weighing left to the officer whose own department held the record.

The objection, and it is the objection the critics actually make when they are precise, is to the method of reconciliation rather than to reconciliation itself. Harmonising two rights that the Constitution Bench in Agarwal had already held capable of structured coexistence could have taken several forms: refining the tests in clause (j), adding a definition of personal information, directing officers to the proportionality factors the privacy jurisprudence supplies, or exempting genuinely private categories by name. The method chosen removed the balancing machinery altogether and left a categorical threshold. It is also worth recording what the expert process had earlier advised: the Group of Experts on Privacy chaired by Justice A. P. Shah, reporting in 2012, and the committee chaired by Justice B. N. Srikrishna, reporting in 2018 with a draft data protection bill, both treated the coexistence of transparency law and privacy law as a problem to be solved by calibrated exceptions and balancing, and neither proposed deleting the public interest test from the RTI Act as the means of reconciliation. As for the assurance about other laws, it is accurate and narrower than it sounds. Disclosure duties imposed by other statutes, a regulator’s disclosure regime, a court’s direction, a scheme’s own transparency rules, are untouched by Section 44(3). What the assurance does not reach is the disclosure duty that the RTI Act itself imposes, because that is precisely the duty the substitution redefined; to say that other laws are unaffected is to say nothing about the one law that was amended. Applicants who relied on the RTI Act to obtain personal information with a public dimension were not relying on other laws, and it is their position that the trail must record.

The narrowing at work: requests before and after

The practical consequence of the substitution is best seen at the level of the requests applicants actually file, because clause (j) was never an abstraction: it was the exemption invoked against the everyday traffic of the Act, and its transformation changes outcomes in recognisable categories of case. Consider the categories in turn, as analysis of how the two texts process the same request rather than as a report of decided cases under a clause whose meaning is still being worked out.

A request for the list of beneficiaries of a welfare scheme, the muster rolls of a works programme, or the names of pensioners paid from a public fund seeks information that relates entirely to identifiable individuals. Under the former clause, such requests were the Act’s classic success story, because the information’s relationship to public activity, the expenditure of public money on named recipients, took it outside the exemption, and social audits in several states were built on exactly such disclosures, exposing ghost beneficiaries and payments to the dead and the fictitious. Under the substituted clause, the authority’s first answer writes itself: the information relates to personal information, and the clause asks nothing further. The applicant’s case must be rebuilt from Section 8(2) and from the argument that a payment record is a record of public expenditure before it is a record of a person, arguments a commission may accept, but arguments the applicant must construct and carry rather than entitlements the clause conferred.

A request for the assets and liabilities declared by a public servant, or for the disciplinary record of an officer against whom a complaint is pending, sits at the point where the former clause did its most contested work. Deshpande had held such material to be personal information disclosable only where a larger public interest was shown, a restrictive reading that applicants criticised even before the substitution, but the route it left open, show the larger public interest, was a route stated in the clause itself, available in every case and applied by commissions across the country. The substituted clause closes that stated route. The officer’s service record relates to personal information; the inquiry ends where it begins, unless the authority exercises its discretion under Section 8(2), and the decided cases that taught applicants how to frame the larger public interest have become authorities on a repealed test, useful as evidence of what the balance once required and of nothing else.

A request for the marks obtained by candidates in a public examination, the merit list of a public recruitment, or the identity of the contractor and the bid amounts in a public tender follows the same migration. Each record relates to individuals; each was disclosable, in whole or in severed part, because its relationship to a public process, a competition for public posts or public money, satisfied the former clause’s first test or its override. Each must be fought for on discretion and characterisation. The pattern across categories is the substitution’s real signature, and it is more precise than the slogan that the Act has been gutted or the counter-slogan that nothing changed: information that is about the state remains as available as it was, and information that is about a person in their dealing with the state, which is where accountability for money, posts and favours is actually traced, has moved from a qualified entitlement administered through stated tests to a refusal administered through a threshold, defeasible only by a discretion the applicant does not control. Which of those two descriptions dominates the Act’s future depends on how often the traffic of requests falls in the second class, and anyone who has read a commission’s cause list knows that the second class is most of the traffic.

The substituted clause inside Section 8: what surrounds it

An exemption clause does not operate alone, and part of the substitution’s significance lies in how it sits among neighbours drafted on the opposite philosophy. Section 8(1) contains ten clauses, and several of them carry internal qualifications of the kind clause (j) lost. Clause (a) exempts information whose disclosure would prejudicially affect the sovereignty, integrity and security of India, a harm test built into the exemption’s own words. Clause (d) protects commercial confidence and competitive position, subject to satisfaction that the larger public interest warrants disclosure, an internal override of the same family as the one clause (j) surrendered. Clause (e) protects fiduciary information on the same overridden pattern. The Act’s drafting convention, in other words, was to state an exemption and then state its public interest release valve, and clause (j) as enacted followed the convention twice over, with its override and its proviso. The substituted clause is the only provision in the sub-section that exempts a subject matter outright, by relation alone, with no stated harm and no stated release. Whatever commissions and courts make of it, they will be construing a clause that is anomalous within its own section, and anomalies of that kind invite either a reading down, to restore consistency with the section’s scheme, or a reading at face value, on the principle that the legislature is taken to have meant the contrast it enacted. That fork, and not any abstract debate about privacy, is what the run of commission orders under the substituted clause will actually decide.

Two further features of the surrounding scheme sharpen the point. Section 8(3) provides that certain exemptions, those in clauses (a), (c) and (i), fall away twenty years after the event to which the information relates, a sunset that expresses the 2005 judgment that secrecy decays. Clause (j) was never within that sunset, before the substitution or after it, so personal information is exempt in principle for as long as the record exists, and the substitution changed the clause’s width without any corresponding attention to its duration. And the proactive disclosure duty in Section 4 continues to require authorities to publish, without any request, categories of information that include the particulars of recipients of concessions, permits and authorisations and the beneficiaries of subsidy programmes, which is to say that the Act still commands authorities to place classes of personal information in the public domain unasked, while the substituted clause entitles them to refuse some of the same classes when a citizen asks. Whether Section 4 publication survives a substituted clause (j) objection is a question the two provisions will have to settle between them in practice, and the tension is recorded here because it is the kind of internal inconsistency that a committee stage, reading the host bill’s amendment section against the amended Act as a whole, exists to catch.

One more category of request shows the clause’s literal reach at its widest, and it is the category commentary noticed first. The substituted clause does not distinguish between information about a stranger and information about the applicant. Under the former clause, a person seeking their own service record, their own examination answer book, which the Supreme Court had held in the Aditya Bandopadhyay decision of 2011 to be obtainable by the examinee, or their own file in a departmental proceeding could be refused only through the tests, and the tests were rarely answerable against the data subject themselves. Under the substituted clause, the authority’s literal answer to the data subject is the same as its answer to anyone, because the clause asks only whether the information relates to personal information, and a person’s own record relates to personal information if anything does. Commissions may well read the clause down for own-information requests, or route them through Section 8(2), and the DPDP Act separately gives a data principal rights of access against data fiduciaries, but the routes are discretionary or external where the former clause was direct. A substitution defended as a protection for the citizen’s privacy that begins by empowering the state to refuse the citizen their own file is a substitution whose drafting outran its rationale, and that sentence, more than any rhetoric about killing the Act, is the criticism its defenders have yet to answer.

What an applicant can still argue

A narrowing is not a repeal, and the advocate’s question after the substitution is a practical one: what routes survive for the applicant whose request meets a refusal under the substituted clause. The general position set out in this section is legal information about the structure of the Act, not advice on any individual application, and an applicant with a live refusal should take advice on their own facts, because appeal periods under the Act are short and strictly applied.

The principal surviving route is Section 8(2), which the substitution did not amend. Section 8(2) provides that, notwithstanding anything in the Official Secrets Act 1923 and notwithstanding any of the exemptions permitted under Section 8(1), a public authority may allow access to information if the public interest in disclosure outweighs the harm to the protected interests. Every word of that provision does work the deleted override used to do, and two of its features define its limits. It operates across the whole of Section 8(1), so it reaches the substituted clause (j) as it reaches every other clause. And it is discretionary where the deleted override was, within clause (j), an operative part of the exemption’s own definition: the old clause asked whether the larger public interest justified disclosure as part of deciding whether the exemption applied at all, while Section 8(2) permits disclosure where the balance favours it, a permission addressed to the authority’s judgment rather than a condition the applicant can insist has been satisfied. An applicant’s argument under Section 8(2) must therefore be built to move a discretion: identify the public interest with specificity, the expenditure audited, the eligibility verified, the misconduct evidenced, rather than asserting public interest in the abstract, and put the argument on the record at the application stage and again on first appeal, so that the Commission on second appeal is reviewing a refusal that confronted the argument rather than one that never heard it.

Why is Section 8(2) a weaker safety net than the override it replaced?

Because the deleted override operated inside the exemption itself, so the public interest inquiry had to be performed before clause (j) could be held to apply, while Section 8(2) permits a public authority to disclose where the balance favours it. A duty to consider is a right the applicant can enforce; a discretion is a favour the applicant must persuade.

Section 8(2) also retains its full reach. It opens with a non-obstante clause that overrides the Official Secrets Act 1923 as well as every exemption in Section 8(1), and the DPDP Act left it exactly as the Parliament of 2005 enacted it, so the one public interest provision that survives in the Act is also the one with the widest field of operation. Its limit is built into its verb: the authority may disclose, so the section supplies a discretion to be persuaded and reviewed rather than an entitlement the applicant can demand.

Two further arguments survive alongside Section 8(2). Severability under Section 10 requires that where a record contains both exempt and non-exempt material, the part that can reasonably be severed and does not fall within an exemption must be disclosed, so a refusal that treats a whole file as personal information because parts of it relate to an individual is vulnerable to the demand that the authority sever and release the rest. And the threshold of the substituted clause is itself an argument: the clause exempts information which relates to personal information, and information about a scheme, a payment, a decision or a process does not become personal information merely because a name appears in it, so the characterisation of the record is contestable in every case. The decisions built on the former clause, Deshpande on service records and Canara Bank v C. S. Shyam on postings and transfers, were decided under a text that no longer exists and cannot be applied as if its tests survived; their value is as a record of how the balance was struck when a balance existed, and as a measure, in argument before a commission, of what the substitution removed. How commissions read the substituted clause across the run of cases, whether they import a public activity inquiry into the word “relates”, or treat the threshold as self-executing, is not settled, and this article does not pretend otherwise: the practical meaning of the narrowing will be fixed by that body of orders and by the constitutional challenge, not by the text alone.

Two procedural features of the unamended Act also remain in the applicant’s hands, and they matter more under a categorical clause than they did under a balanced one. Section 19(5) places the onus of justifying a denial on the Public Information Officer who denied the request, in any appeal proceedings, so an authority that refuses under the substituted clause must still justify the refusal before the Commission, and a justification that does no more than recite the clause’s words, without addressing the record’s character, the severable parts or the Section 8(2) balance the applicant has raised, is a justification the Commission is entitled to find incomplete. Section 11, in turn, requires that where information relating to a third party has been treated as confidential by that party, the officer must give the third party notice and an opportunity to be heard before deciding, which inserts a hearing, and with it a reasoned order, into a class of cases that includes many personal information disputes. Neither provision restores the deleted tests, and neither should be oversold. What they preserve is a procedural discipline around the refusal, a burden on the refuser and a hearing before the refusal hardens, and in a regime whose exemption is a threshold, the discipline around the threshold is where applicants will win or lose.

The laying control: Parliament’s remaining handle on the rules

One objection to the critics’ account of 2019 is sometimes made and should be answered on the record, because it identifies the safeguard the design does retain. Rules made by the Central Government under the RTI Act are subject to the laying requirement in Section 29: they must be placed before each House of Parliament, and if both Houses agree to modify or annul a rule within the laying period, the rule takes effect as modified or not at all. The 2019 Rules were laid accordingly. Parliament, on this argument, did not surrender the tenure question; it retained a veto over the executive’s answer to it, and the three-year term subsists with Parliament’s acquiescence recorded in the only form the procedure requires.

The argument is accurate about the procedure and weak about the institution, for reasons that the practice of delegated legislation in India has made familiar. Laying is a negative control: a rule takes effect on notification and stands unless Parliament affirmatively acts against it, which inverts the burden that primary legislation imposes, where nothing stands until Parliament affirmatively enacts it. Annulment requires a motion, time in a crowded calendar, and agreement in both Houses, and motions to annul or modify rules are among the rarest events in parliamentary practice, while the committees on subordinate legislation that examine rules for excess of mandate report on vires and drafting rather than on policy merits like the proper length of a commissioner’s term. The 2019 amendment’s defenders are therefore right that the form of parliamentary control survived and wrong about what the form delivers: a tenure that Parliament must organise itself to veto is, in the working constitution, a tenure the executive sets. The contrast the trail preserves is between a five-year term that could be displaced only by a bill the government had to pilot through two Houses, as the 2019 Bill itself had to be piloted and nearly was not, and a three-year term that could be displaced only by a parliamentary initiative no government need ever fear. The safeguard against disadvantageous variation protects the sitting commissioner; the laying procedure protects, in theory, the institution; and the distance between those two protections and the fixed statutory term they replaced is the measure of the first narrowing.

The challenge in court and the campaign for repeal

The substituted clause operates under a constitutional cloud, and the state of that litigation must be described structurally, as this series describes every pending matter: what has been filed, what the court has done, and what remains open, with no prediction of the result. Petitions challenging Section 44(3), together with challenges to other provisions of the DPDP Act including its provisions on the Data Protection Board and on government information-gathering powers, are pending before the Supreme Court. The petitioners include journalists and press bodies, whose case is that investigative reporting depends on processing personal data and on obtaining disclosures about public activity, and that the substitution converts a calibrated exemption into a blanket bar that restricts the freedom of speech and expression under Article 19(1)(a) beyond what Article 19(2) permits. The Court has issued notice to the Union, has declined to stay the substitution, describing the questions as complex and unsuited to interim resolution, and has referred the questions raised to a larger bench, so the substituted clause governs applications while its validity awaits authoritative determination. Senior counsel for the petitioners have directed the Court to its own Constitution Bench reasoning in Agarwal, where the balance between information and privacy was treated as a structured exercise rather than a categorical choice, and the Union’s defence rests on the harmonisation argument set out in the previous section. That is the whole of what can responsibly be said: the clause is law, the clause is challenged, and the challenge is undecided.

Alongside the litigation runs a legislative campaign, because a provision enacted by Parliament can also be repealed by Parliament, and the opposition has pursued that route in terms. In March 2025, Jairam Ramesh, a Congress member of the Rajya Sabha, wrote to the Minister for Electronics and Information Technology urging that the government pause, review and repeal Section 44(3), arguing that the existing clause had contained adequate protection for privacy and that the substitution would destroy the working of the RTI Act. In April 2025, opposition members of Parliament of the INDIA bloc carried the demand into a joint letter to the same minister, signed by 120 members according to the press reports of its release, among them the Leader of the Opposition in the Lok Sabha, Rahul Gandhi, and senior members from the Congress, the Dravida Munnetra Kazhagam, the Shiv Sena (Uddhav Balasaheb Thackeray), the Samajwadi Party, the Communist Party of India (Marxist) and the Rashtriya Janata Dal. The joint letter’s argument tracks the analysis of the deleted text: that the substitution exempts personal information even where it pertains to public functions and public money, and that it removes the proviso which tied the citizen’s entitlement to Parliament’s own. In July 2025, Justice A. P. Shah, a former Chief Justice of the Delhi High Court and a former Chairperson of the Law Commission of India, addressed an open letter to the Attorney General for India, R. Venkataramani, written after the ministry administering the DPDP Act had sought the Attorney General’s opinion on whether that Act undermines the RTI Act. Justice Shah’s letter argued that the changes destroy a delicate equilibrium the 2005 Act had struck, that a narrowly tailored exemption had been replaced by an overbroad one, and that the amendment should be rolled back. A reader will note the symmetry, because it is the kind of detail a trail exists to catch: the jurist who had chaired the 2012 expert group that recommended balancing as the reconciliation between privacy and transparency was, thirteen years on, writing to the government’s own law officer to protest a reconciliation achieved by deleting the balance. The government has not accepted the repeal demand, and the provision’s commencement in November 2025 proceeded while both the campaign and the litigation were live. For any reader who has to keep the evidentiary record of this fight straight, the applications, orders, letters and judgments that accumulate around a contested provision are exactly the material that has to be stored so that no part of it can be rewritten later, which is why journalists working on accountability stories keep their source files in an encrypted workspace, following the practices journalists use to protect sources with VaultBook.

The arguments the larger bench will hear

The pending challenge is described structurally elsewhere in this article, as the rules of this series require for undecided litigation, but the arguments on each side can be stated in their strongest form without predicting which will prevail, and stating them completes the trail because the challenge is, functionally, the scrutiny the provision did not receive in Parliament, conducted in a forum where both sides must answer questions.

The petitioners’ case begins from the lineage of the right itself. The Supreme Court held in State of Uttar Pradesh v Raj Narain in 1975, and again in S. P. Gupta in 1981, that the citizen’s right to know is implicit in the freedom of speech and expression under Article 19(1)(a), and the disclosure cases of the following decades, including the electoral disclosure decisions in the Association for Democratic Reforms line, treated access to information about public functioning as a constituent of democratic choice rather than a statutory favour. On that foundation, a restriction of the right must answer to Article 19(2) and to the proportionality discipline the Court has built around it since Puttaswamy: a legitimate aim, a rational connection, necessity, and a balance that leaves the right more than a residue. The substitution, the petitioners argue, fails at necessity and balance together, because the former clause demonstrates on its face that a less restrictive instrument was available and in use, one which protected privacy against unwarranted invasion while preserving disclosure where public activity or larger public interest justified it. A legislature that replaces a calibrated instrument with a categorical one must explain what the calibrated instrument got wrong that only categoricity can cure, and the explanation offered, harmonisation with privacy, is said to collapse on inspection, because Puttaswamy itself protected privacy through proportionality and the Constitution Bench in Agarwal had already performed the harmonisation the amendment claims to seek. The petitioners add the press dimension: investigative journalism’s reliance on disclosures about named officials and beneficiaries means the clause burdens newsgathering directly, and the DPDP Act’s absence of a journalistic exemption for its own processing obligations compounds the burden on the same activity from the other side.

The Union’s case, at its strongest, is a case about legislative choice among reasonable designs. Privacy is a fundamental right; the state holds personal data on a scale no private actor matches; the former clause entrusted the balance to the very officers whose departments held the records, producing inconsistency that the petitioners’ own tradition of criticism has documented for years; and a legislature is not constitutionally confined to the least speech-restrictive design imaginable, provided the design it chooses is rational, non-arbitrary and directed to a legitimate end. The substituted clause, on this view, is not a secrecy provision but a classification: it assigns personal information, as a class, to the protected side of the line, leaving the rest of the Act, including Section 8(2)’s discretion and every other clause’s operation, exactly where it was. Courts ordinarily defer to such classifications in social and regulatory legislation, and a bench asked to restore a deleted balancing test is being asked to re-enact a policy choice the legislature deliberately unmade. The Union can add that the sky has not fallen in the manner predicted: the Act’s machinery runs, disclosures continue across the vast run of records that are not personal information, and the petitioners’ strongest examples are capable of being met, case by case, by a Section 8(2) discretion exercised as Parliament intended.

Between those cases sits the question the bench itself flagged in referring the matter onward, whether a blanket prohibition on a class of disclosure, defined by an undefined phrase, can be reconciled with the structured approach the Court’s own precedents prescribe. That question is open, it is genuinely difficult, and the trail’s discipline is to leave it exactly there.

The cumulative arc: drift from two directions

An amendment trail earns its name by reading the whole sequence as one movement, and the RTI Act’s sequence has a shape that neither amendment shows alone. The Act of 2005 was built on a distrust of executive discretion at two points: it fixed the adjudicators’ terms in the statute so the executive could not reach them, and it qualified the privacy exemption with tests and an override so the executive’s officers could not expand it by interpretation. The amendments of 2019 and 2023 each relaxed one of those restraints, by opposite techniques. The 2019 amendment enlarged executive discretion over the institution while leaving the text of the rights untouched, and it did so openly, in a bill that named its subject and survived a contested division. The 2023 substitution enlarged executive discretion over the exemption while leaving the institution untouched, and it did so collaterally, inside a host statute whose debate belonged to another subject. Between them they exhibit the two standard modes by which a statute drifts: amendment of its machinery by a government that controls the legislative timetable, and amendment of its substance by a provision that never has to face the constituency the statute serves. Which mode a given legal system should fear more is a question about scrutiny, and it is the question the comparison table in this article is built to answer. The deeper institutional safeguard against the collateral mode is the committee system itself, since a bill examined clause by clause by members charged with one subject is a bill whose stray provisions get noticed, which is why the working of parliamentary committees in bill scrutiny is the necessary companion to any amendment trail that ends, as this one twice does, with a committee stage that never happened.

The comparison also disciplines a claim that this cluster makes elsewhere. Transparency statutes and anti-corruption statutes are both said to decay in office, and the comparison of the RTI Act with the Lokpal takes up which of the two designs is the more vulnerable to being weakened; this article’s contribution to that comparison is the observation that the RTI Act’s vulnerability proved to be located not in its celebrated operative sections, which no amendment has touched, but in its service conditions and in a single clause of its exemption list, the two places a reader scanning for repeal would never think to look. A statute’s strength, the trail suggests, should be measured at its least watched provisions, because that is where amendment is cheapest.

What the trail predicts follows from the same logic. The 2019 mechanism is reusable: any future adjustment of the commissions’ terms, in either direction, can be made by notification, and the precedent that tenure lives in rules will not be unmade except by a Parliament willing to legislate the numbers back into the Act. The 2023 mechanism is equally reusable and more general, because every statute in the Indian code is exposed to amendment from inside any later statute whose drafters choose to include it, and the only systematic defences are committee scrutiny of host bills, a press and a profession that read amendment sections, and courts willing to ask whether a substitution of this width was a proportionate reconciliation of two rights. The repeal campaign and the pending challenge are, in that sense, attempts to rebuild after the event the scrutiny that the legislative process did not supply before it.

Reversibility: why the two narrowings are not equally permanent

A final comparison between the two narrowings concerns their reversibility, and it cuts against the assumption that the more debated change is the more durable one. The 2019 relocation is, in form, the easiest kind of change to undo that Indian public law knows. Because the term and the salary live in rules, a future Central Government that wished to restore a five-year term and a higher salary could do so by amending the 2019 Rules, an executive act requiring no bill, no majority and no debate, in the same Gazette series in which the three-year term was announced. The delegation that critics condemned as executive capture is indifferent to the direction of travel: it empowers a government to shorten commissioners’ terms and it equally empowers a successor to lengthen them, and the critics’ true objection, properly stated, is not that three years is tyrannical but that the question is one no executive should hold in its gift at all, in either direction. Whether any future government will use the power benignly is a speculation this article does not entertain; the structural record is that the power exists, is held by the respondent in the commissions’ cases, and has been used once, downward.

The 2023 substitution is locked in by the opposite mechanism. It is primary legislation, a section of an Act of Parliament, and it can be undone only by another Act of Parliament, which is what the repeal campaign demands and what no government has offered, or by a court striking it down, which is what the pending challenge seeks and what no court has yet done. There is no rule to amend and no notification to withdraw, because the change was made at the highest level of legislative form, by the route that attracted the least legislative attention. The inversion is complete and worth stating plainly, since it is the trail’s most portable lesson: the narrowing that Parliament debated for a week can be reversed by a signature, and the narrowing that Parliament barely noticed can be reversed only by Parliament or a Constitution Bench. Scrutiny and durability, which a healthy legislative process would align, point in opposite directions on this record, and a reader who learns to ask of any future amendment not only what it changes but how it is entrenched, and how it can be unmade, has learned the habit this series exists to teach.

Verdict

Tested provision by provision, the popular claim that the RTI Act has been killed does not survive its own statute book: the right in Section 3, the request machinery, the appeal ladder, the penalty, the removal protections and most of the exemption scheme stand exactly where the Parliament of 2005 left them, and applications continue to be filed, answered, appealed and penalised at scale. The counter-claim, that nothing of substance changed, fails faster, and it fails on documents anyone can read: a commissioner’s term that was five years in an Act is three years in a rule, a salary that was an Election Commissioner’s is a figure in a notification, and an exemption clause that required three questions to be asked before refusal can be satisfied by one. The accurate verdict is the narrower one this trail has argued throughout. The Act was narrowed twice, at two different load-bearing points, by two different legislative techniques, and the second narrowing, which removed the balancing tests and the Parliament proviso from the exemption applicants meet most often, is the more consequential change to the Act’s working meaning, while also being the change that received the least examination before it became law. That inversion, the depth of a change varying inversely with the scrutiny it received, is the collateral amendment’s signature, and learning to detect it in a bill is the skill this article exists to teach: read the amendment section of every statute, because the statute you care about may be amended there, by a sub-section, in a week when everyone is watching something else.

Frequently Asked Questions

Q: Which sections did the RTI Amendment Act 2019 amend?

The Right to Information (Amendment) Act 2019, Act 24 of 2019, amended three sections of the Right to Information Act 2005. Section 13, which governs the term of office and the salaries and conditions of service of the Chief Information Commissioner and the Information Commissioners, was amended so that the term and the salary ceased to be fixed in the Act and became matters to be prescribed by the Central Government. Section 16, the corresponding provision for the State Chief Information Commissioner and the State Information Commissioners, received the identical amendment. Section 27, the Central Government’s rule-making power, was amended so that the newly prescribable matters, the term of office and the salaries, allowances and other terms of service of the central and state commissioners, could in fact be prescribed by rule. The Act received assent on 1 August 2019 and came into force on 24 October 2019.

Q: Why was the tenure of Information Commissioners reduced?

Strictly, the amending Act of 2019 did not itself reduce the tenure. It removed the five-year term from Sections 13 and 16 of the RTI Act and replaced it with a term to be prescribed by the Central Government, and the reduction happened when the government exercised that power: the Right to Information (Term of Office, Salaries, Allowances and Other Terms and Conditions of Service) Rules 2019, notified on 24 October 2019, prescribed a term of three years for the Chief Information Commissioner, the Information Commissioners and their state counterparts. Critics of the amendment treat the two steps as one event, because the delegation was the decision that made the reduction possible without further parliamentary involvement, and because the rules were notified on the very day the amendment commenced. Defenders reply that three years is a respectable term found elsewhere in the statute book. Both descriptions are accurate about different halves of the mechanism: Parliament surrendered the number, and the executive then chose a smaller one.

Q: What justification did the government give for the 2019 RTI amendment?

The government’s stated justification was a classification argument, presented in both Houses by Dr Jitendra Singh, the Minister of State for Personnel, Public Grievances and Pensions. The Election Commission, he argued, is a constitutional body created by Article 324 of the Constitution, while the Information Commissions are statutory bodies created by an Act of Parliament, so equating the service conditions of information commissioners with those of election commissioners was an anomaly in the original drafting that the amendment corrected. The minister assured Parliament that the amendment carried no design to curtail the commissions’ autonomy, that it would streamline the working of the RTI Act, and that the prescribing power would not be used arbitrarily. The government’s supporting statements added a uniformity rationale: service conditions across India’s many statutory commissions and tribunals vary widely, and the Information Commissions were being brought within a prescribable framework like the general run of such bodies, rather than left benchmarked to a constitutional office.

Q: Did the RTI Amendment 2019 go to a select committee?

No. The opposition in the Rajya Sabha moved that the Bill be referred to a select committee of the House, and the motion was defeated on a division, with 117 members voting against reference and 75 members voting for it, after a turbulent sitting in which opposition members protested about the conduct of the voting and the Congress walked out before the Bill’s final passage. The Bill had also not been examined by the department-related standing committee on personnel, public grievances, law and justice, which is the other route by which a bill of this kind ordinarily receives clause-by-clause examination and takes evidence from outside witnesses. The Bill was therefore enacted on the strength of floor debate alone: introduced in the Lok Sabha on 19 July 2019, passed there on 22 July, passed by the Rajya Sabha on 25 July, and assented to on 1 August 2019, thirteen days from introduction to enactment.

Q: How did the Data Protection Act change the RTI Act?

The Digital Personal Data Protection Act 2023, Act 22 of 2023, changed the RTI Act through a single provision in its amendment section. Section 44(3) of the DPDP Act substituted clause (j) of Section 8(1) of the RTI Act, the personal information exemption. The former clause exempted personal information whose disclosure had no relationship to any public activity or interest, or which would cause unwarranted invasion of the privacy of the individual, unless the Public Information Officer or the appellate authority was satisfied that the larger public interest justified disclosure, and a proviso added that information which cannot be denied to Parliament or a State Legislature shall not be denied to any person. The substituted clause reads, in full, information which relates to personal information. The substitution deleted the public activity test, the unwarranted invasion test, the internal public interest override and the Parliament proviso at once, and it was brought into force on 13 November 2025 by notification G.S.R. 843(E).

Q: What was the public interest override in Section 8(1)(j)?

The public interest override was a provision inside the former clause (j) of Section 8(1) of the RTI Act that allowed disclosure of personal information despite the exemption where the Central or State Public Information Officer, or the appellate authority on appeal, was satisfied that the larger public interest justified the disclosure of such information. Its position inside the clause mattered as much as its content. Because the override formed part of the exemption’s own definition, an officer deciding a request involving personal information had to work through it as part of deciding whether the exemption applied at all, asking whether the information lacked a relationship to public activity or interest, whether disclosure would work an unwarranted invasion of privacy, and whether the larger public interest nevertheless justified release. The Supreme Court’s decisions under the former clause, including Deshpande in 2012 and the Constitution Bench in Agarwal in 2019, applied that structure. Section 44(3) of the DPDP Act deleted the override when it substituted the clause in 2025.

Q: Is the amendment to Section 8(1)(j) being challenged in court?

Yes. Petitions challenging Section 44(3) of the Digital Personal Data Protection Act 2023, together with challenges to other provisions of that Act, are pending before the Supreme Court. The petitioners include journalists and press bodies, and their case is that substituting a clause which barred disclosure only after a balancing exercise with a clause that bars disclosure of information relating to personal information, without a public interest release, restricts the right to information under Article 19(1)(a) beyond what the Constitution permits, and removes the structured balancing the Constitution Bench endorsed in the Agarwal decision. The Supreme Court has issued notice to the Union of India, has declined to stay the substitution, and has referred the questions to a larger bench. The substituted clause therefore operates as law while its validity is under challenge, and the outcome cannot be predicted from the state of the proceedings. The challenge is directed at the provision’s validity, and no interim protection against its application has been granted.

Q: Has the RTI Act been weakened or merely adjusted?

The provision-by-provision record supports weakened, in two specific respects, over adjusted. The 2019 amendment weakened the institution: a five-year term and Election Commission salary parity, fixed in the statute, became a term and salary prescribed by the executive, and the rules then set three years, a change no mere adjustment produces, because it relocates the power to decide for every future holder of the office. The 2023 substitution weakened the right at its most used point: clause (j) of Section 8(1) lost the public activity test, the unwarranted invasion test, its internal public interest override and the Parliament proviso, leaving a threshold that asks only whether information relates to personal information. Against that, the adjustment case can fairly note what survives intact: the right in Section 3, the request and appeal machinery, the penalties, the removal protections and most of the exemption scheme. The accurate description is a working statute narrowed at its adjudicators’ tenure and at its commonest exemption, which is more than adjustment and less than repeal.

Q: When did the RTI Amendment Act 2019 come into force?

The Right to Information (Amendment) Act 2019 came into force on 24 October 2019. The Bill had been introduced in the Lok Sabha on 19 July 2019, passed by the Lok Sabha on 22 July 2019 and by the Rajya Sabha on 25 July 2019, and the Act, numbered Act 24 of 2019, received the assent of the President on 1 August 2019. Commencement was left to notification, and the Central Government brought the Act into force on 24 October 2019. The same day, it notified the Right to Information (Term of Office, Salaries, Allowances and Other Terms and Conditions of Service) Rules 2019, which exercised the new prescribing power and set the commissioners’ term at three years and their pay at Rs 2,50,000 a month for a Chief Information Commissioner and Rs 2,25,000 a month for other commissioners, at the Centre and in the states alike. Enactment and commencement were therefore separate events nearly three months apart, and the operative change in commissioners’ terms dates from the October notification, not from the August assent.

Q: What salary do the 2019 Rules prescribe for Information Commissioners?

The Right to Information (Term of Office, Salaries, Allowances and Other Terms and Conditions of Service) Rules 2019 prescribe pay of Rs 2,50,000 fixed per month for the Chief Information Commissioner and pay of Rs 2,25,000 fixed per month for Information Commissioners, and the same two figures for the State Chief Information Commissioner and the State Information Commissioners respectively. The Rules add dearness allowance and other facilities aligned to posts carrying the same pay, and they set off pension against pay where a commissioner draws a pension from earlier government service. For comparison, the position the Rules replaced was salary by equivalence under Sections 13 and 16 of the RTI Act: the Chief Information Commissioner’s salary was the same as the Chief Election Commissioner’s, and an Information Commissioner’s the same as an Election Commissioner’s, figures pegged by law to the salary of a judge of the Supreme Court, which stands at Rs 2,50,000 a month. The Chief’s prescribed figure therefore matched the former parity figure in cash, while the Information Commissioners’ figure was set Rs 25,000 a month below it, and both figures lost the self-revising character that equivalence had given them.

Q: How quickly did Parliament pass the RTI Amendment Bill 2019?

The Bill completed its passage through Parliament in seven days and became law thirteen days after it was introduced. It was introduced in the Lok Sabha on 19 July 2019, passed by the Lok Sabha on 22 July 2019, taken up and passed by the Rajya Sabha on 25 July 2019, and received the President’s assent on 1 August 2019 as Act 24 of 2019. Within that week, the Rajya Sabha also voted on and defeated, by 117 votes to 75, the opposition’s motion to refer the Bill to a select committee, and the Bill was never examined by the department-related standing committee either. Reporting of the government’s file notings published subsequently indicated that the presiding officers of both Houses had been asked to expedite the Bill and to relax the ordinary notice requirements for its introduction. The Act was then brought into force later, on 24 October 2019, so the three-month interval between assent and commencement was taken up by the preparation of the rules that would prescribe the new term and salaries, rather than by any further parliamentary stage.

Q: Did the 2019 amendment change what the Information Commissions can order?

No. The 2019 amendment changed the conditions on which commissioners hold office and left their powers untouched. The Central and State Information Commissions retain the powers the RTI Act gave them in 2005: the power under Section 18 to inquire into complaints with the powers of a civil court, the power under Section 19 on second appeal to direct disclosure, to require a public authority to take steps to secure compliance, and to award compensation in the cases the section specifies, and the power under Section 20 to impose a penalty of two hundred and fifty rupees for each day of default, up to a maximum of twenty five thousand rupees, on a Public Information Officer who has refused an application, delayed without reasonable cause, or knowingly given incorrect or misleading information, together with the power to recommend disciplinary action. The removal protections in Sections 14 and 17, the bar on reappointment of commissioners, and the age ceiling of sixty five also survived the amendment. What changed was who sets the commissioners’ term and salary, and for how long the first set of rules set them, three years instead of the statutory five.

Q: When did the substituted Section 8(1)(j) take effect?

The substituted clause (j) of Section 8(1) of the RTI Act took effect on 13 November 2025. The substitution was enacted by Section 44(3) of the Digital Personal Data Protection Act 2023, which received assent on 11 August 2023, but like many provisions of that Act it awaited commencement by notification, and during the interval the former clause, with its public activity test, its unwarranted invasion test, its public interest override and its Parliament proviso, continued to govern applications. On 13 November 2025, the Central Government issued notification G.S.R. 843(E), the same exercise that constituted the Data Protection Board and brought the Digital Personal Data Protection Rules 2025 into operation, and brought Section 44(3) into force with immediate effect. From that date, the exemption reads, in full, information which relates to personal information. Most of the DPDP Act’s own compliance obligations were phased in later on the notified schedule, so the amendment to the RTI Act was among the earliest parts of the data protection statute to operate in practice.

Q: What was the Parliament proviso that the substitution deleted from Section 8(1)(j)?

The Parliament proviso was the concluding proviso to the former clause (j) of Section 8(1) of the RTI Act, and it provided that information which cannot be denied to Parliament or a State Legislature shall not be denied to any person. Its effect was to set the legislature’s entitlement as the floor of the citizen’s entitlement: a public authority refusing a request for personal information had to reckon with the question whether the same information could be withheld from Parliament or a state legislature, in a question hour, a committee proceeding or a call for records, and if it could not be withheld there, the proviso forbade withholding it from the applicant. The proviso rested on the theory of responsible government, that the executive answers to the legislature and through it to the citizen, and it functioned in practice as an argument of last resort for applicants who had lost the balancing inside the clause. Section 44(3) of the Digital Personal Data Protection Act 2023 deleted the proviso along with the rest of the former clause when it substituted clause (j) with the words information which relates to personal information, and nothing in the substituted text replaces it.

Q: Can personal information still be disclosed in the public interest after the substitution?

Yes, but through a weaker and discretionary route. Section 8(2) of the RTI Act, which the DPDP Act did not amend, provides that notwithstanding anything in the Official Secrets Act 1923 and notwithstanding any of the exemptions in Section 8(1), a public authority may allow access to information if the public interest in disclosure outweighs the harm to the protected interests. That provision reaches every clause of Section 8(1), including the substituted clause (j), so a public authority confronted with information that relates to personal information retains a statutory power to disclose it where the public interest balance favours disclosure. The difference from the former clause is the quality of the power. The former clause (j) built the larger public interest inquiry into the exemption itself, so an applicant could insist that the inquiry be performed before the exemption was held to apply. Section 8(2) confers a discretion, expressed by the word may, so the applicant’s task is to persuade the authority, and on appeal the Commission, that the discretion should be exercised in favour of disclosure on the facts, by identifying the public interest with specificity rather than asserting it in general terms.

Q: What is a collateral amendment and why does it matter?

A collateral amendment is a provision inside a statute that amends a different statute on a subject outside the host statute’s own field, usually placed in the host’s closing sections among its repeals and amendments and drafted in substitution or omission formulae. Section 44(3) of the Digital Personal Data Protection Act 2023 is the worked example: a sub-section in a data protection statute that rewrote Section 8(1)(j) of the Right to Information Act, deleting its balancing tests and its Parliament proviso. The technique matters because of the scrutiny asymmetry it creates. A direct amending bill announces its target in its title, goes to the committee responsible for that subject, alerts the affected constituency and is debated as what it is. A collateral provision travels under the host bill’s title, is examined, if at all, by the wrong committee with the wrong witnesses, and is covered by the press as part of the host’s story. The legal effect is identical in both cases, so the only defences against a consequential change arriving unexamined are the habits of reading amendment sections, comparing a bill’s amendment schedule against its stated objects, and treating any substitution that deletes overrides and provisos as substantive.

Q: Who has campaigned for the repeal of Section 44(3) of the DPDP Act?

The repeal demand has been carried by opposition parliamentarians and a former Chief Justice, in three documented interventions. In March 2025, Jairam Ramesh, a Congress member of the Rajya Sabha, wrote to the Minister for Electronics and Information Technology urging the government to pause, review and repeal the section, arguing that the former clause had protected privacy adequately. In April 2025, opposition members of the INDIA bloc sent a joint letter to the same minister, signed by 120 members according to the press reports of its release, among them the Leader of the Opposition in the Lok Sabha, Rahul Gandhi, and members of the Congress, the Dravida Munnetra Kazhagam, the Shiv Sena (Uddhav Balasaheb Thackeray), the Samajwadi Party, the Communist Party of India (Marxist) and the Rashtriya Janata Dal, demanding repeal on the ground that the substitution exempts personal information even where it concerns public functions and public money. In July 2025, Justice A. P. Shah, a former Chief Justice of the Delhi High Court and a former Chairperson of the Law Commission of India, addressed an open letter to the Attorney General for India, R. Venkataramani, arguing that the amendment destroyed a delicate equilibrium and should be rolled back.

Q: Did the 2019 amendment also cover State Information Commissioners?

Yes. The 2019 amendment amended Section 16 of the RTI Act, which governs the State Chief Information Commissioner and the State Information Commissioners, in the same terms as Section 13 for the central commissioners: the statutory term of five years and the salary equivalence, with an Election Commissioner for the State Chief and the Chief Secretary of the state government for State Information Commissioners, were replaced by a term and salary to be prescribed by the Central Government. Section 27 was amended so that the prescribing power extends to the state commissioners, and the 2019 Rules accordingly prescribe a three-year term and pay of Rs 2,50,000 a month for a State Chief Information Commissioner and Rs 2,25,000 a month for State Information Commissioners. This central prescription for state offices drew a distinct federal objection in the debate, pressed by P. Chidambaram among others: a state government constitutes its State Commission and its members are appointed through the statutory committee the Act prescribes for the state, so the transfer of their tenure and pay to Central Government rules placed a state institution’s conditions of service in the hands of a different government from the one that constitutes and funds it.

Q: Can the government change the commissioners’ term again without another Act of Parliament?

Yes. That is the structural consequence of the 2019 amendment. Because Sections 13 and 16, as amended, provide that commissioners hold office for such term as may be prescribed by the Central Government, and because Section 27 confers the prescribing power, the term is set by the Right to Information (Term of Office, Salaries, Allowances and Other Terms and Conditions of Service) Rules 2019, which set it at three years. The Central Government can amend those rules by notification and prescribe a different term, longer or shorter, without introducing a bill, securing a majority or facing a debate in either House. The surviving controls are procedural and weak in practice: rules must be laid before Parliament under Section 29 of the Act, and both Houses may modify or annul them, but annulment motions are rare and a rule stands unless Parliament acts. One protection does bind the executive: the terms of a sitting commissioner cannot be varied to that commissioner’s disadvantage after appointment, so a change operates on future appointments rather than on incumbents. Before 2019, by contrast, the five-year term stood in the Act itself, and changing it required the legislative process the 2019 Bill itself had to undergo.

Q: What should an applicant argue when information is refused as personal information?

An applicant refused under the substituted clause (j) has four arguments to build and should make them early; this is general information about the Act’s structure, not advice on an individual case, and appeal periods under the Act are short. First, characterisation: the clause exempts information which relates to personal information, and a record of a decision, payment, process or scheme does not become personal information merely because a name appears in it, so the authority can be pressed to identify what in the record is personal. Secondly, severability: Section 10 requires disclosure of any severable part of a record that falls outside an exemption, so a blanket refusal of a whole file is open to challenge. Thirdly, Section 8(2): the authority retains a discretion to disclose where the public interest in disclosure outweighs the harm to the protected interests, identifying that public interest with specificity: the public money traced, the eligibility verified, the misconduct evidenced. Fourthly, procedure: on appeal the onus of justifying the denial lies on the Public Information Officer under Section 19(5), and where a third party treated it as confidential, Section 11 entitles that party to a hearing before the decision hardens.