AFSPA presents a puzzle that recurs across the north-east and baffles readers who assume that a state’s own government controls the law operating on its own soil. A chief minister tells a legislative assembly that his government wants the Armed Forces (Special Powers) Act, 1958 withdrawn. The assembly passes a resolution, sometimes unanimously, demanding repeal. The demand is reported, debated, and endorsed by parties that disagree about everything else. And the Act stays in force, its disturbed area notifications renewed for another six months, because the decision was never the state’s to take. The statute that produces this result is short, and the argument about it is usually aimed at the wrong section. The fierce public controversy concentrates on Section 4, the provision that lets a soldier fire on a citizen. The provision that actually governs the politics, the federalism, and the grievance is Section 3, the provision that decides where Section 4 is allowed to operate at all. Until a reader understands who holds the pen that signs a disturbed area declaration, and whose consent that signature does and does not require, every other question about AFSPA, its powers, its immunity clause, its court cases, and its future, will be argued at cross purposes.

The claim of this article can be stated in one sentence. Section 3 is the real AFSPA. The powers in Section 4 exist only inside territory covered by a Section 3 declaration, the protection in Section 6 matters only where those powers have been used, and because Parliament gave the declaration power to the Central Government as well as to the Governor and the Administrator, acting without any requirement of state concurrence, the deepest argument about this statute is a federalism argument wearing a human rights costume. Both arguments are real. The human rights argument concerns what soldiers may do and what happens when they exceed it. The federalism argument concerns who decides that a part of a state should be governed, for policing purposes, under an exceptional central statute at all. Conflating the two is the reason the public debate circles for decades without resolving: repeal advocates argue about Section 4 abuses to an audience whose constitutional objection is Section 3, and defenders answer the Section 3 objection with Section 4 necessities, and neither side ever has to answer the other’s actual point. This article separates the two arguments, gives each its strongest form, and then shows where the record supports a verdict.

Different readers need different instruments from the same statute, and the article is organised so that each can extract theirs without wading through the others’ material. The legislator or the legislative staffer will find the consent question and the menu of amendments stated at the level of drafting, because the live choices in Parliament are choices about Section 3’s defaults. The advocate will find the operative law stated as the courts have conditioned it, the inquiry right of 2016, the sanction gate, the handover discipline, and the forums in which each is enforced, because a client’s case under this Act is won or lost on the conditioned law and not on the bare text. The researcher will find the record assembled in one place, the extension and contraction of the map, the committee reports, the sanction episodes, and the comparative frame, because the scholarly argument about AFSPA is an argument from the record and the record is scattered. The citizen of a notified district, whose questions are the most concrete of all, is answered in the situations worked through later in the article and in the questions at its close. What unites the four is the discipline of beginning with the declaration: for the legislator it is the clause to amend, for the advocate the fact to establish, for the researcher the document to find, and for the citizen the signature that decides which legal world her street inhabits.

AFSPA and the disturbed area: Section 3 is the real statute

Section 3 of the Act is brief enough to summarise in a sentence, and everything in this series of disputes flows from its allocation of a single power. Where the Governor of a state, the Administrator of a union territory, or the Central Government is of the opinion that the whole or any part of the state or territory is in such a disturbed or dangerous condition that the use of the armed forces in aid of the civil power is necessary, that authority may, by notification in the Official Gazette, declare the whole or that part to be a disturbed area. Three features of this drafting deserve slow attention, because each one carries a consequence that the later history merely unfolded. The first is the threshold: the statute does not require an insurgency, a rebellion, or any defined level of violence. It requires an opinion, formed by the declaring authority, that conditions are disturbed or dangerous enough to make military aid to the civil power necessary. The second is the identity of the opinion holder: the power is conferred in parallel on the state’s constitutional head, the union territory’s administrator, and the Central Government itself, with no hierarchy among them and no requirement that any one of them obtain the agreement of another. The third is what the section does not contain: there is no time limit in the text, no requirement of legislative approval, no duty to record reasons, no consultation clause, and no sunset. Every limit that exists in the working law of declarations, the six month review cycle, the expectation of consultation, the minimum force discipline, was supplied later by the Supreme Court reading discipline into a silent text, or by executive practice, and not by Parliament in 1958.

Does the Centre need a state’s permission to declare its territory disturbed?

No. Section 3 confers the declaration power on the Central Government in its own right, and the Act nowhere requires the concurrence, consent, or prior agreement of the state government concerned. The Supreme Court confirmed this reading in 1997, adding only that the Centre should ordinarily consult the state before acting on its own motion.

Consultation, in this scheme, is not consent: a state that is consulted and objects has been heard, not obeyed, and a declaration made over its objection is valid law. That single design choice explains the recurring political theatre of AFSPA. When a declaration is issued by the Governor, it is in constitutional form the act of the state itself, because a Governor ordinarily acts on the aid and advice of the elected council of ministers, so a state government that dislikes a Governor’s notification can ordinarily procure its withdrawal through the same channel that produced it. The position is transformed when the notification is the Centre’s. A state government confronting a central declaration has no statutory lever of its own: it cannot revoke another government’s notification, it cannot legislate the Act out of its territory because the Act is a central law operating in a field of Union competence, and its remedies are political representation, public pressure, and negotiation with the Ministry of Home Affairs, the ministry that administers the Act. This is why state demands for withdrawal so often take the form of assembly resolutions and chief ministerial letters rather than legal instruments. The resolution is not a failure of nerve; it is the only instrument the statute leaves in state hands. It is also why the periodic extensions that attract the least attention are the most legally significant events in the Act’s life: every six month renewal of a central notification is a fresh exercise of Union power over the internal policing map of a state, taken in Delhi, in a field, public order, that the Constitution’s Seventh Schedule assigns in the first instance to the states.

The declaration is also the Act’s only gateway, and this is the analytical point on which the whole statute turns. Section 4 powers do not float over India waiting to be used. They are dormant in any place, however violent, until a Section 3 notification covers that place, and they lapse in any place, however violent, when the notification covering it is allowed to expire or is withdrawn. A police station area removed from a notification is, for AFSPA purposes, ordinary India again, where soldiers have no more authority than any citizen and where an arrest by the Army would be tested against the ordinary criminal law. The map of disturbed areas is therefore not background information about the Act; it is the Act, in operative form. Two further consequences follow. Because the gateway is territorial and notification based, the real administrative law of AFSPA is a law of notifications, boundaries, and renewals, conducted largely through gazette publications that few citizens ever read. And because the declaring authorities are plural, a single state’s map can be a palimpsest of notifications from different sources and different decades, so that answering the apparently simple question of whether AFSPA applies in a particular district requires identifying which notification, issued by whom, covers it, and when that notification is next due to expire.

The table below is this article’s findable artifact: the declaration and consent map, setting out each declaring authority, the reach of its power, the consent position, the review that the Supreme Court’s 1997 judgment attached to it, and what the record shows in practice.

Declaring authority	Who may declare, and over what territory	State concurrence required?	Review under the 1997 judgment	What the practical record shows
The Governor of a state	The whole or any part of the state, by notification in the Official Gazette, acting ordinarily on the aid and advice of the state council of ministers	Not applicable in the central sense: the declaration is constitutionally the state’s own act, so the state government can ordinarily secure its withdrawal through the same channel	Declaration to be for a limited duration, with periodic review before the expiry of six months	State notifications have been the route for several withdrawals, including the lifting of the notification from the Imphal municipal area in 2004
The Administrator of a union territory	The whole or any part of the union territory, by notification in the Official Gazette	No: a union territory is administered by the Centre, so the concurrence question does not arise in federal form	The same limited duration and six month periodic review discipline applies	Manipur was a union territory when the Act was first extended to it, and the administrator route mattered in the Act’s early geography
The Central Government	The whole or any part of any state or union territory to which the Act extends, by its own notification, a power conferred by the amendment of 1972	No: the Act requires no state concurrence, and the 1997 judgment asks for consultation before a suo motu declaration, not consent	Limited duration, periodic review before six months expire, and consultation with the state government before a declaration made on the Centre’s own motion	Central notifications and their six month renewals have carried the Act through most of its long tenures, and a state assembly resolution cannot revoke them
The state government as such	Holds no declaration power of its own under the Act apart from the Governor acting on its advice; it may request, recommend, protest, and resolve	Its agreement is politically weighty and usually sought in practice, but it is not a statutory condition of a central declaration	No review function is assigned to it by the Act or the judgment	Assembly resolutions demanding repeal, including the unanimous Nagaland resolution of December 2021, are political instruments addressed to the Centre, not legal acts that alter the notification map

One further feature of Section 3 completes the legal picture, and it concerns the courts.

Can a disturbed area declaration be challenged in court?

Yes, within limits. A declaration rests on the declaring authority’s opinion, and the courts will not substitute their own assessment of security conditions, but a declaration can be struck down if it is mala fide, rests on no relevant material, or is one no reasonable authority could have made on the material available. The 1997 judgment proceeded on that footing.

That footing reflects the general doctrine for opinion based powers in Indian administrative law, which are neither unreviewable nor reviewable on appeal. The 1997 judgment treated the declaration as open to judicial scrutiny on those ordinary limited grounds while declining to convert the courts into appellate authorities over security assessments. In practice, challenges to declarations as such have rarely succeeded, because the material underlying a security assessment is seldom fully before the court and the margin the doctrine concedes is wide; the successful litigation under the Act has overwhelmingly concerned acts done under declarations rather than the declarations themselves, the excessive force, the delayed handover, the refused sanction. The consequence is an accountability structure tilted toward the incident and away from the regime: a single unlawful killing can be condemned and compensated while the declaration that placed the force in the village runs on unexamined. Reformers who speak of making the six month review substantive usually mean, in legal terms, an attempt to correct exactly this tilt, by attaching to the renewal a record that a court, a legislature, or the public could actually test.

The powers that follow: Sections 4, 5 and 6

Once a declaration is in force, Section 4 confers the operative powers, and it confers them remarkably low in the chain of command. The section speaks to any commissioned officer, warrant officer, non-commissioned officer, or any other person of equivalent rank in the armed forces. In plain terms, the powers that dominate the controversy can be exercised not only by senior commanders but by the junior leaders who actually stand at checkpoints and lead patrols, which is a deliberate legislative choice about where discretion should sit in a counter-insurgency environment, and also the reason the safeguards question cannot be answered by pointing to the seniority of the decision maker. Four clusters of power are granted. First, after giving such due warning as the officer considers necessary, the officer may fire upon or otherwise use force, even to the causing of death, against a person acting in contravention of a law or order in force in the disturbed area that prohibits the assembly of five or more persons or the carrying of weapons, firearms, ammunition, or explosive substances. Second, the officer may destroy arms dumps, fortified positions, and structures used as training camps or shelters from which armed attacks are made or likely to be made. Third, the officer may arrest without warrant a person who has committed a cognisable offence or against whom a reasonable suspicion exists that he has committed or is about to commit one, using such force as is needed for the arrest. Fourth, the officer may enter and search any premises without warrant to make such an arrest or to recover persons, arms, or explosives believed to be held there.

What can a soldier lawfully do in a disturbed area that a police officer cannot?

Four things, in combination: after due warning, the soldier may use force up to and including lethal force against a person breaching a prohibitory order on assemblies or weapons; may arrest on reasonable suspicion without a warrant; may enter and search premises without a warrant; and may destroy arms dumps and fortified positions.

A police officer, by contrast, works under the ordinary criminal procedure, where lethal force is confined to narrow defences and arrest and search are structured by warrant and record requirements that Section 4 displaces. Two features of Section 4 are persistently misdescribed in public argument, and precision about them serves both sides of the debate. The lethal force power is not a free standing licence to shoot: it is tied to contravention of a prohibitory order of the specified kinds, it requires the officer’s opinion that the action is necessary for the maintenance of public order, and it is preceded by due warning. Critics answer, correctly, that these conditions are thin protection in practice because the opinion is the officer’s own, the warning standard is what he considers necessary, and the prohibitory orders themselves are executive instruments. Defenders answer, also correctly, that a power to fire only after warning and only against persons defying notified prohibitions is a different thing from the caricature of an unrestricted power to kill, and that the conditions have real disciplinary content inside a force that trains to them. What neither side can honestly deny is the structural fact: Parliament chose to write the conditions as matters of the officer’s judgment rather than as matters of prior authorisation, and that choice, made in Section 4, is what the judicial safeguards of later decades have had to work upon.

Section 5 is the shortest of the operative provisions and, in the daily friction between the Army and the citizen, among the most important. A person arrested under the Act must be made over to the officer in charge of the nearest police station with the least possible delay, together with a report of the circumstances occasioning the arrest. The design is a handover design: the armed forces detain, the civil police receive, and from the moment of handover the ordinary criminal process, with its magistrate production requirements and its record, is supposed to take over. The Supreme Court in 1997 read the handover duty together with the constitutional requirement that an arrested person be produced before a magistrate within twenty four hours, and the Army’s own list of do’s and don’ts carries the same discipline. The friction arises in the gap between the two clocks: the statute’s clock runs in the language of least possible delay, which is a standard rather than a number, and in remote terrain the distance to the nearest police station is itself a variable. Families experience the handover interval as a period in which a person is held by a force that is not required to maintain the paperwork of a police station, and the habeas corpus petitions that punctuate the Act’s history are usually filed in precisely that interval. The provision is therefore best understood as a bridge that the Act builds between military detention and civil process, and the litigation record is largely a record of disputes about how long that bridge is permitted to be.

What happens to a person arrested by the armed forces under AFSPA?

The arrested person must be handed over to the officer in charge of the nearest police station with the least possible delay, together with a report of the circumstances of the arrest. From that handover the ordinary criminal process governs, and the Constitution’s twenty four hour rule for production before a magistrate frames the custody that follows.

Section 6 completes the architecture, and it is the provision around which the accountability argument turns. It provides that no prosecution, suit, or other legal proceeding shall be instituted against any person in respect of anything done or purported to be done in exercise of the powers conferred by the Act, except with the previous sanction of the Central Government. Three points of precision matter here. The bar is a bar on the institution of proceedings, not a declaration that the acts themselves are lawful: it is a gate at the courthouse door, held by the executive, and not a finding on the merits. The gatekeeper is the Central Government in every case, including cases arising inside a state whose own government may favour prosecution, which is a second federalism sting inside the Act: a state police force may investigate, a state government may recommend, and the file still travels to Delhi for the decision that determines whether a court may hear the matter at all. And the bar covers civil suits and other legal proceedings as well as prosecutions, so its practical reach extends beyond criminal trials to the compensation claims that are often the only remedy a family is realistically pursuing. The section’s defenders locate it in a familiar family of sanction provisions across Indian law, which require government approval before public servants are prosecuted for acts done in office, on the reasoning that operational decisions taken in good faith must not be re-litigated by prosecution at the instance of every complainant. Its critics answer that a sanction provision is only as defensible as the record of its exercise, and that the record, examined later in this article, is the weakest point in the entire statutory scheme.

The renewal machine: how a temporary declaration becomes a standing regime

Because the Act contains no time limit, the entire discipline of duration in AFSPA’s working life comes from the interaction of the 1997 judgment with executive practice, and that interaction has settled into a machine whose operations are worth describing precisely. A disturbed area notification is issued for a stated period, in practice six months, matching the review interval the Supreme Court prescribed. As the period runs out, the security establishment prepares an assessment: the Ministry of Home Affairs, which administers the Act, takes inputs from the state government, the state police, the Army and the central armed police forces deployed in the theatre, and the intelligence agencies, and a decision is taken whether the notification should be extended for a further six months, reduced in area, or allowed to lapse. The decision issues as a fresh gazette notification, and the cycle repeats. Nothing in this process requires a debate in any legislature, a vote, a published statement of reasons addressed to the public, or the agreement of the affected state, and the notifications themselves are terse instruments that record the declaration without narrating the assessment behind it. A citizen who wishes to know why her district remains disturbed for the fortieth consecutive six month period will find the notification, the fact of renewal, and nothing else: the reasoning lives in files to which the accountability channels described in this article have only imperfect access.

The machine’s political economy explains its outputs. Renewal is the path of least resistance for every participant in the assessment except the state government that wants withdrawal. For the security establishment, recommending withdrawal carries an asymmetric risk: if violence follows, the recommendation is blamed, whereas renewal that proves unnecessary carries no comparable cost. For the Ministry, a renewal preserves an option and a lapse surrenders one. For the state government, the assessment is one input among several, and where the operative notification is the Centre’s, the state’s view can be recorded and overridden without any public accounting of the disagreement. The result is the pattern the map shows: renewals are routine and rarely newsworthy, reductions come in negotiated tranches attached to political moments, peace accords, changes of government, the aftermath of a scandal, and full withdrawals come when an insurgency is plainly finished. Defenders of the machine answer that counter-insurgency assessments are genuinely technical, that six month review is exactly the periodicity the Supreme Court prescribed, and that the reductions of 2015, 2018, and 2022 show the machine does move in both directions. Critics answer that a review conducted by the beneficiaries of continuation, on unpublished criteria, with the burden silently resting on withdrawal, is review in form and renewal in substance, and that the 1997 judgment’s discipline has been complied with in the letter by the very regularity that empties it. Both descriptions fit the same facts; the disagreement is over what the word review should be taken to require, and it is the live form of the declaration debate.

Reading a notification: the anatomy of the operative document

Since the Act lives in its notifications, a reader equipped to read one is equipped to follow the regime in her own district, and the document repays the small effort involved. A disturbed area notification is a short gazette instrument: it cites Section 3 as its source, names the declaring authority, describes the territory it covers, and states the period for which the declaration runs. The territorial description is the operative heart, and it is drawn in the units of police administration, whole districts, named subdivisions, or the areas falling within the jurisdiction of named police stations, which is why modern maps of the Act look mottled: the legal geography follows the police map, not the physical one, and two adjoining villages can stand on opposite sides of the line if their police stations differ. The period stated, conventionally six months from a named date, is the reader’s cue to the renewal cycle: each notification carries its own expiry, and the question whether AFSPA applies in a place on a given day is answered by finding the notification in force for that police station’s area and checking that its period has not run out. Extension notifications are drafted as fresh declarations rather than amendments, which is the administrative reflection of the 1997 judgment’s insistence that each period rest on a fresh review. Two practical consequences follow for anyone studying the regime. The gazette, and the home department records of the state and the Ministry, are the primary sources, and secondary maps in newspapers and reports lag behind them, sometimes by a full renewal cycle. And the unit of analysis for every serious question, where the Act applies, since when, on whose signature, and until when, is the individual notification, so that a state level statement about AFSPA, whether celebratory or condemnatory, is always a summary that the underlying documents may complicate.

Section 4 on the ground: orders, suspicion, and the texture of the powers

The abstractions of Section 4 acquire a specific texture when they are followed into the street, the checkpoint, and the cordon and search operation, and the texture explains grievances that the statutory text alone does not predict. The lethal force power is keyed to prohibitory orders, and the prohibitory orders in a disturbed district are the standing instruments of public order administration: orders restricting the assembly of five or more persons, orders restricting the carrying of arms, and curfew type orders during operations, issued by the district administration under the public order provisions of the criminal procedure. The five person threshold inside Section 4 echoes the definition of unlawful assembly in the penal law, and its practical consequence is that a gathering which would elsewhere be a matter for police dispersal becomes, in a disturbed area after warning, a lawful occasion for military force, which is why the legality of the underlying prohibitory order, its promulgation, its communication, and its exceptions, matters so much in litigation and is so seldom examined in political debate. The arrest power runs on reasonable suspicion, a standard the officer forms on the spot, and it is not accompanied by the graduated safeguards the courts have built around police arrest under the ordinary code, because the arresting authority is not the police and the Act’s own safeguard is the handover duty of Section 5. The search power is exercised most consequentially in the cordon and search operation, in which a village or locality is sealed and houses are searched for arms and sheltered persons, an operation type the Army’s instructions surround with requirements about the presence of civil officials and the treatment of women and property, and which generates a large share of the complaints that later travel the accountability chain, most of them turning not on the power’s existence but on the manner of its use.

The definitions in Section 2 quietly decide several of these practical questions. The Act defines the armed forces as the military forces and the air forces operating as land forces, and any other armed forces of the Union so operating, which is the textual route by which central paramilitary forces deployed in aid of civil power, pre-eminently the Assam Rifles in the north-east, exercise Section 4 powers, and by which the state police, not being armed forces of the Union, do not. The same section defines the disturbed area by reference to the Section 3 notification, closing the circle: the powers exist where the declaration exists, for the persons the definition covers, against the conduct the prohibitory orders forbid. A reader who holds those three definitional wires, territory, personnel, and triggering conduct, can analyse almost any reported incident under the Act by asking the three questions the definitions pose: was the place notified, was the actor within the definition, and was there a prohibitory order or a cognisable suspicion of the kind the section specifies. Incidents in which the legal argument later collapsed, for one side or the other, usually collapse on one of those three wires, and the discipline of asking is worth more than most of the slogans the debate supplies.

Three situations the consent map decides

The value of the declaration and consent map can be tested against the three situations in which readers actually meet this statute. In the first, a state government elected on a promise to remove AFSPA discovers that the operative notifications for its districts are central notifications. Its legal department confirms what the map shows: it may withdraw any Governor’s notification through the Governor acting on its advice, it may recommend withdrawal of the central notifications, and it may make the political case in every available forum, but the notifications it most wants gone are not its instruments to revoke, and its practical strategy becomes negotiation over tranches, a district here and a set of police stations there, of exactly the kind the reductions of 2022 represent. In the second situation, a family whose relative has been killed in an operation asks what the law offers. The map and the case law give a structured answer: an investigation must occur and cannot be blocked by the Act, a magisterial or judicial inquiry may examine the death, the human rights commission may seek a report from the Central Government under the special procedure for the armed forces, compensation may be sought from the courts on constitutional principles, but a criminal prosecution of the personnel involved begins only if the Central Government grants previous sanction, and the family’s realistic litigation is therefore likely to be about compelling investigation and contesting a sanction refusal rather than about a trial. In the third situation, a state government facing sudden violence asks for the Army’s deployment and discovers the converse of the consent problem: it can request deployment in aid of civil power through the ordinary channels, but the special powers arrive only with a Section 3 declaration, which it may issue through the Governor or ask the Centre to issue, and the choice of channel has consequences, because a Governor’s notification remains within the state’s power to withdraw while a central notification does not. In each situation the decisive fact is the same, and it is a Section 3 fact: who signed, and who can unsign. That is the sense in which the declaration provision, and not the firing provision, is the real AFSPA, and it is the first thing a legislator, an advocate, or a citizen should establish in any concrete controversy under this Act.

A colonial draft: from the Ordinance of 1942 to the Act of 1958

Many Indian statutes are called colonial as a loose epithet for anything old and coercive. In AFSPA’s case the description is a drafting fact that can be traced document by document, and the trace matters because it explains the statute’s shape. On 15 August 1942, in the opening days of the Quit India Movement, the colonial government promulgated the Armed Forces (Special Powers) Ordinance, conferring on officers of the armed forces special powers in aid of the civil authority, including the power to use force against those defying prohibitory orders and a protection against proceedings without official sanction. The family resemblance to the 1958 Act is not a matter of spirit or tendency; it is visible in the architecture. A triggering condition defined by executive opinion, powers conferred on officers down the chain of command, force up to causing death after warning, arrest and search without warrant, and a sanction gate for proceedings: the 1942 instrument and the 1958 Act share a skeleton. Independent India first reused the pattern in the disturbed areas ordinances of 1947, issued for Bengal, Assam, East Bengal, and the United Provinces amid the violence of Partition, and a general Armed Forces (Special Powers) Act followed in 1948 before being repealed in 1957 as unsuited to the constitutional order then taking shape. The 1958 revival was therefore not an invention but a return to a stored design, taken down from the shelf when a particular emergency seemed to require it.

The emergency was the Naga insurgency. The Naga Hills, then a district of Assam, saw an organised movement for separation led by the Naga National Council, a boycott of the first general elections, and, by 1956, a parallel administration proclaimed by the Naga nationalist leadership. Assam responded first with its own instruments: the Assam Maintenance of Public Order legislation applied to the hills, and then the Assam Disturbed Areas Act of 1955, a state statute that provided a legal frame for the paramilitary and armed police operations then being conducted. When those instruments proved insufficient for a campaign that the state judged to be beyond police capacity, the Union stepped in with a central law for the Army. The President promulgated the Armed Forces (Assam and Manipur) Special Powers Ordinance on 22 May 1958, and Parliament replaced it with the Armed Forces (Assam and Manipur) Special Powers Act, which received assent on 11 September 1958. The original title records the original geography: the composite state of Assam, which then included the Naga Hills and much of what later became the separate states of the north-east, and Manipur, then a union territory. The Act was a short statute for a defined theatre, presented to Parliament as a response to a specific armed challenge in a specific pair of territories. Nothing in its passage suggested a general internal security code for the Republic, and the distance between that modest origin and the statute’s later career is one of the central facts any honest account of AFSPA has to hold in view: a theatre specific emergency measure became, by extension and renewal rather than by fresh parliamentary decision, one of the longest lived exceptional laws in Indian history.

The parliamentary passage of 1958 fixed expectations that the later career of the Act disappointed, and the disappointment is part of the historical record rather than a retrospective invention. The bill was defended by the government of the day as an exceptional and localised measure, directed at a specific breakdown of order in the Naga Hills and the adjoining areas, in a debate in which members warned against writing emergency powers into ordinary law and against the precedent of governing Indian citizens through a statute modelled on a colonial ordinance. The government’s assurances carried the bill, and the assurances are worth recovering precisely because they state the basis on which Parliament consented: a defined theatre, a specific armed challenge, and an instrument expected to retire with the emergency that justified it. The general Armed Forces (Special Powers) Act of 1948 had in fact been allowed to lapse and be repealed within a decade, which made the expectation of impermanence reasonable in 1958. What no participant in that debate legislated for was the combination that produced the modern Act: territorial extension by the amendment of 1972, a declaration power for the Centre itself, and a renewal practice that converted a theatre measure into a standing feature of governance for several states. Historians of the statute treat the 1958 debates as the baseline of a familiar Indian legislative pattern, in which the exceptional law is passed on assurances of brevity that the text does not contain and the text, not the assurances, is what the future inherits.

One Act, many maps: expansion, contraction, and the six month cycle

The geography of the Act after 1958 is a history of the north-east’s political reorganisation and of its violence in equal measure. As the composite Assam was divided into new states, the Act followed the map. The decisive statutory step came with the amendment of 1972, which extended the Act across the north-eastern states, Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, and Tripura, and, of equal constitutional significance, conferred the Section 3 declaration power on the Central Government itself. Before 1972 the Centre directed operations through the state and territorial authorities; after 1972 it held a pen of its own. The amendment is therefore the true hinge of the federalism story: the consent problem that dominates the modern debate was not present in the Act as passed in 1958 and entered it fourteen years later, by amendment, without anything like the public attention that the original enactment received. The Act also travelled outside the north-east in adapted forms. Punjab’s insurgency years were governed by a companion statute, the Armed Forces (Punjab and Chandigarh) Special Powers Act of 1983, modelled on the 1958 pattern, which remained in force until 1997, when it was repealed as the Punjab insurgency ended. Jammu and Kashmir received a separate central statute of its own in 1990, the Armed Forces (Jammu and Kashmir) Special Powers Act, 1990, again closely modelled on the 1958 Act and differing from it in arrangement rather than in spirit. A persistent error in public writing treats the 1958 Act as the law operating in Kashmir; it is not, and the distinction matters to any reader trying to identify which sanction decisions, which notifications, and which judgments belong to which statute.

The contraction of the map is as instructive as its expansion, because it shows what the end of an AFSPA tenure actually looks like in administrative practice. Tripura was brought under the Act in 1997, at the height of its insurgency, and the state government withdrew the disturbed area notification in May 2015 after eighteen years, judging that the level of violence no longer justified the exceptional regime; the withdrawal was a state act, taken on a state assessment, and it has held. Meghalaya’s exposure was always partial: the Act operated in a belt of territory along the Assam border, twenty kilometres deep, rather than across the whole state, and the Centre removed the notification from that belt with effect from 1 April 2018, after twenty seven years, as the security assessment changed. Mizoram’s insurgency ended with the peace accord of 1986, among the most durable settlements in the region’s history, and the state passed out of the Act’s practical geography as its notifications lapsed and the insurgency that had justified them was settled politically. Manipur illustrates the granular character of the modern map: after the protests of 2004, the state government withdrew the disturbed area status from the Imphal municipal area, so that for years the capital’s municipal limits stood outside the Act while the districts around them remained inside it, a single city divided by a notification line. The reductions announced in 2022 continued the same method at larger scale: substantial sets of districts and police station areas in Assam, in Manipur, and in Nagaland were taken out of the notifications, while the Act was retained in the areas the security establishment judged still active, and the retained notifications continued on their six month renewal cycle. In Manipur the violence that began in 2023 rearranged the map again, with the Act retained across the hill districts while the valley police station areas stood outside the notification. The pattern across six decades is consistent: the map contracts police station by police station and district by district, in step with security assessments and political settlements, and it almost never contracts by the dramatic statewide repeal that public debate demands. Whether that pattern represents prudent gradualism or the institutionalisation of permanence is the political question at the heart of the repeal debate, and it is a question about Section 3 practice, not about Section 4 text.

How the Act travels: extension, adaptation, and the sibling statutes

The mechanism by which AFSPA reached new territory is itself instructive, because the Act was not drafted as a national code awaiting application but as a named-territory statute that had to be carried, place by place, by further legislation and amendment. Its original extent was the state of Assam and the union territory of Manipur, and its first journey was to Tripura, carried by an extension Act of 1970 addressed to that territory. The general extension came with the amendment of 1972, which applied the Act across the seven north-eastern states as the region’s state boundaries were redrawn, and at the same time equipped the Central Government with its own declaration power. The pattern after 1972 was therefore dual. Within the north-east, the Act stood available as a standing statute whose operation in any district awaited only a notification. Outside the north-east, where Parliament wished to deploy the same machinery, it legislated afresh: the Armed Forces (Punjab and Chandigarh) Special Powers Act of 1983 for the Punjab theatre, and the Armed Forces (Jammu and Kashmir) Special Powers Act of 1990 for Kashmir, each reproducing the essential architecture of declaration, special powers, handover, and sanction in a self contained code for its own territory. Alongside the central statutes there also subsist state instruments of the same family, of which the Assam Disturbed Areas Act of 1955 is the ancestor: state laws under which a state government may declare areas disturbed for the purposes of its own forces and the forces operating in aid under its direction. A map of special powers in India is therefore a layered map, of a central Act extended state by state, of sibling central Acts for particular theatres, and of state Acts beneath them, and a surprising amount of public confusion, including litigation confusion, comes from attributing the acts of one layer to another. The careful reader of any incident report asks first which statute, of which legislature, supplied the powers being exercised, because the sanction authority, the declaration authority, and even the text of the powers differ at the margins between layers, and arguments won against one layer are routinely and wrongly advertised as won against all.

The extension history also corrects a simplification in the repeal debate, in which AFSPA is spoken of as a single switch that a single decision could turn off. Repeal of the 1958 Act would end the regime in the north-east, but the Punjab precedent shows both the possibility and the pattern of an ending: the 1983 Act was repealed when its insurgency ended, without the sky falling, and the Jammu and Kashmir Act would survive any repeal of the 1958 Act untouched, requiring its own parliamentary decision. Conversely, the existence of sibling statutes shows that the 1958 Act’s survival is not a legislative oversight but a repeated legislative choice: Parliament has returned to this architecture in 1972, in 1983, and in 1990, each time with full knowledge of the controversy the pattern attracts. A reform campaign that addresses only the 1958 text therefore addresses only part of the structure, and the deeper object of criticism, disclosed by the extension record, is a legislative habit: when a theatre of internal violence appears, reach for the stored design, extend it, and let the renewal machine administer the consequences. The habit, rather than any single enactment, is what six decades of criticism have failed to break.

The constitutional setting: public order belongs to the states, until it does not

The federal objection to AFSPA begins with the Seventh Schedule. Public order is Entry 1 of the State List, and police is Entry 2: the ordinary constitutional scheme assigns the maintenance of order inside a state to that state’s government and its police force, answerable to its legislature. A central statute that authorises the Union’s armed forces to exercise police type powers inside a state, on a declaration the Union can issue for itself, sits in obvious tension with that assignment, and the tension was the ground of the principal constitutional attack on the Act. The Union’s answer, which the Supreme Court accepted in 1997, rests on the Union List and on Article 355. Entry 2A of the Union List, inserted by the Forty second Amendment in 1976, gives Parliament power over the deployment of the armed forces of the Union in any state in aid of the civil power, and Article 355 casts a duty on the Union to protect every state against external aggression and internal disturbance and to ensure that the government of every state is carried on in accordance with the Constitution. The Court’s reasoning was that AFSPA is, in pith and substance, a law about deploying the Union’s armed forces in aid of civil power, a Union subject, and not a law on public order as such: the forces act in aid of the civil power, the civil authority is not displaced, and no proclamation of Emergency is required because none of the constitutional consequences of an Emergency, the suspension of rights enforcement, the takeover of state government, follows from a disturbed area declaration. The judgment’s discussion of the distribution of legislative power between the Union and the states remains the standard account of why a statute policing state territory can be a central law.

The honest limits of that answer should also be stated. Entry 2A was itself inserted in 1976, eighteen years after the Act, during the Emergency Parliament, so the competence argument that now sustains AFSPA was consolidated retrospectively around a statute that predates it; the Court dealt with the point by reference to the Union’s defence and armed forces entries as well, but the sequence is a reminder that constitutional doctrine in this area was built around a statute that governments of both parties wished to preserve. More substantially, the aid to civil power formula does increasing work as deployments lengthen. Aid implies a principal and an assistant, a civil authority that requests and directs and a military force that helps; a deployment renewed every six months for decades, in which the armed forces conduct the principal counter-insurgency campaign and the state police operate at its margins, tests the formula’s descriptive honesty even where its legal validity is settled. The federalism critique of AFSPA, at its strongest, is not that the Act is unconstitutional, for the Court has held that it is not. It is that a constitutional scheme which assigns public order to the states has been made to accommodate, through the combination of Entry 2A, Article 355, and a self issued declaration, a parallel and centrally controlled order maintaining apparatus inside consenting or non-consenting states alike, and that the accommodation has no natural stopping point in the text. The statute belongs in this respect to a wider family of exceptional security laws that suspend ordinary process in defined situations, a family whose peacetime member is the preventive detention code, and the family resemblance is instructive: in each case the exceptional power is defensible at the moment of crisis, and in each case the constitutional argument is really about duration, review, and who decides that the exception should continue.A further distinction, elementary but persistently blurred in debate, separates the Act from martial law. The Constitution makes no general provision for the imposition of martial law across a region; Article 34 empowers Parliament to indemnify officials for acts done during martial law where it has been in force, but no Indian government has governed a state under a martial law proclamation of the classical kind, and AFSPA is not one. In a disturbed area the civil courts sit, the state government functions, the ordinary statutes run, and the armed forces act under a parliamentary statute whose every exercise is in principle examinable by the courts that continue to operate around them. The point matters to both sides. To the Act’s defenders it is the answer to the charge of military rule: soldiers in a disturbed area are statutory auxiliaries of a functioning civil order, not its replacement. To its critics it sharpens rather than blunts the complaint, because a regime that keeps every form of civil government intact while hollowing out the substance of civil protection has chosen the hollowing deliberately, provision by provision, and must answer for the choice in a way that a genuine emergency regime, with its sunsets and its parliamentary approvals, is never allowed to escape.

The comparison with President’s Rule under Article 356 explains why governments of all parties reach for AFSPA and leave it in place. A proclamation under Article 356 displaces the state’s elected government, transfers its functions to the Centre, requires parliamentary approval within two months, and runs against strict constitutional time limits and searching judicial review of the kind the Supreme Court developed in the Bommai judgment. A disturbed area declaration under Section 3 does none of this: the state government remains in office, the assembly sits, elections proceed, and the only visible change is that the Union’s armed forces inside the notified areas carry statutory powers and the Centre acquires a renewable foothold in the state’s order maintenance. AFSPA is, in institutional terms, the low cost instrument of central presence in a troubled state: it attracts none of the political price of dismissing a government, it requires no parliamentary moment at all, and it can be calibrated district by district in a way that a constitutional proclamation cannot. The 1997 Court was right in law that this is not an emergency by another name, because the legal incidents of emergency are absent. The political scientist’s observation that it performs part of an emergency’s function, sustained central coercive authority inside a state without the state’s consent and without emergency’s safeguards and sunsets, is also right, and the coexistence of those two truths is the precise location of the AFSPA problem. It is why the declaration power, rather than the firing power, draws the fire of federalists: Section 4 governs what happens in an encounter, but Section 3 governs the constitutional terms on which a part of India is policed, and it governs them indefinitely, at executive discretion, in the space the Constitution’s more demanding instruments were designed to make difficult.

The states answer back: requests, refusals, and the asymmetry in action

The federal argument is sometimes conducted as if the states were uniformly unwilling hosts of a central imposition, and the record is more interesting than that: states have requested the Act, accepted it, administered it, withdrawn it, and raged against it, sometimes in sequence within a single state’s history, and the variety is itself evidence about where the design fails. Assam in November 1990 shows the Centre acting at maximum intensity with the state’s constitutional government displaced: facing the ULFA insurgency, the Union dismissed the state government, imposed President’s Rule, declared the whole of Assam a disturbed area, and launched a full Army operation, a sequence in which the disturbed area declaration was one instrument inside a general central takeover rather than a freestanding deployment. Manipur in 2004 shows the state’s residual power used at its margin: after the protests of that summer, the state government withdrew the disturbed area status from the Imphal municipal area, an act within its competence because the operative instrument was susceptible to state withdrawal, and the capital’s exemption then persisted for years as a standing demonstration that partial withdrawal was administratively possible, while the surrounding districts remained under the Act. Tripura shows the complete arc within one state’s politics: brought under the Act in 1997 when its insurgency was at its height, governed under it for eighteen years, and taken out of it in 2015 by a state government that judged the insurgency finished, with the succeeding government of a different party content to leave the withdrawal undisturbed, a sequence suggesting that where the withdrawal power is effectively in state hands, withdrawal decisions track security conditions rather than party. Jammu and Kashmir shows the refusal pattern under the sister statute of 1990: in 2011 the chief minister announced his government’s intention to revoke the Act from several districts where the Army’s presence had thinned, the proposal met open institutional resistance from the Army, and the revocation did not take place, leaving in place a declaration whose continuation no elected government of the state had chosen. Nagaland in 2021 completes the set with the unanimous assembly resolution that changed nothing on the notification map. Arranged together, the episodes isolate the variable that matters. Where the state holds the pen, declarations rise and fall with conditions and politics, as federal theory predicts. Where the Centre holds the pen, declarations persist through changes of state government, unanimous resolutions, and completed police investigations, and the state’s constitutionally primary subject, public order, is administered on a timetable set elsewhere.

The institutional referees of Indian federalism noticed the problem early, in general form. The Sarkaria Commission on Centre State relations, reporting in 1988, addressed the deployment of Union forces in the states and recommended that such deployment should ordinarily be made at the request of the state or in consultation with it, that the forces should act in aid of and in consultation with the state administration, and that resort to unilateral deployment should be exceptional and explained. The Commission was writing about deployment generally rather than about AFSPA’s declaration mechanism, and its recommendations were never converted into law, but its framework states the norm against which the Section 3 design is measured in federal argument: consultation as the rule, consent as the ordinary expectation, unilateral action as the justified exception. Measured against that norm, the Act as amended in 1972 inverts the defaults. Consultation is a judicially encouraged courtesy, consent is legally unnecessary, and unilateral continuation is the administrative routine. Defenders of the design reply that insurgency is precisely the exceptional case the norm contemplates, and that a state government captured by local political pressures may be the worst judge of whether an armed challenge to the Union should be met by the Union’s forces. The reply is a serious one about the north-east’s early decades, in which state administrations were weak, penetrated, or themselves in negotiation with armed movements. It is a weaker reply about the Act’s sixth decade, in which the states concerned are established governments with their own police forces, assembly mandates, and, in the Nagaland case, a completed investigation by their own officers. The federalism debate about AFSPA is at bottom a debate about whether an exception designed for the first situation should be governed by rules written for it after the second has arrived.

The comparative check: how other democracies arm their soldiers at home

A brief comparison disciplines both sides of the argument, because it shows that the problem AFSPA solves is real and general while its particular solution is a choice. In the United Kingdom, from which the 1942 Ordinance’s drafters came, military aid to the civil power historically rested on the soldier’s position as a citizen: a soldier called out to assist the civil authority possessed in principle no powers beyond those of any citizen, acting under the direction of the civil magistrate, which is one reason the long Northern Ireland deployment generated its own exceptional statutory overlays, most notoriously the Civil Authorities (Special Powers) Act of 1922 in Northern Ireland, a special powers regime of arrest, detention, and delegated regulation making that survived for half a century and was dismantled in the early 1970s as part of the reckoning with that conflict. In the United States, the Posse Comitatus Act of 1878 states the opposite default for a federation: the federal Army may not be used as a domestic police force except where the Constitution or an Act of Congress expressly authorises it, so that every domestic military role requires a specific legislative key of the kind AFSPA represents, and the authorisation debate happens in the legislature at the point of grant. The Indian design differs from both in where it locates the continuing decision. It is neither the British model of civil direction with citizen powers, nor the American model of a statutory prohibition lifted case by case by the legislature, nor even its own 1958 model of a theatre specific enactment. It is a standing legislative key, granted once, whose use in any particular district is thereafter an executive act of declaration and six monthly renewal, with the legislature’s role exhausted at enactment. Comparativists who study special powers regimes note that this is the configuration most prone to permanence, because the decision to continue is made by the institution that operates the regime, at intervals short enough to be routine and long enough to avoid scrutiny, and the Indian record under AFSPA is frequently cited as the demonstration. The comparison does not decide the Indian question, since neither Britain’s Irish experience nor America’s federal structure maps onto the north-east’s borders and insurgencies. It does identify the feature of the Indian design that carries the explanatory weight, and it is once again a Section 3 feature: not the existence of special powers, which democracies grant in emergencies of many kinds, but the location and the longevity of the power to switch them on.

What the Supreme Court did in 1997: validity, with conditions attached

The constitutional fate of the Act was settled in Naga People’s Movement of Human Rights v Union of India, decided on 27 November 1997 by a five judge Constitution Bench, which heard the challenges to the Act together and dismissed them. The petitioners’ case had two limbs. The first was competence: public order and police being state subjects, Parliament could not enact a law of this character for state territory. The second was character: a statute that suspends the ordinary protections of criminal process across whole regions for indefinite periods, without a proclamation of Emergency and without the safeguards that attend one, was said to be an emergency achieved by ordinary legislation, a colourable device. The Court rejected both limbs, for the Entry 2A and Article 355 reasons already described, and upheld Sections 3 to 6 against challenges under Articles 14, 19, and 21. Had the judgment stopped there, it would be remembered as a simple validation. It did not stop there, and the second half of the judgment is the part that governs the Act’s daily operation. Reading the statute as a whole and against the Constitution, the Court attached operative conditions to the powers it upheld, conditions that bind because the Court declared them to be the law of the Act’s exercise.

The conditions can be grouped under four heads. Duration and review: a declaration under Section 3 is to be for a limited duration, and there must be periodic review of the declaration before the expiry of six months, so that the six month extension cycle familiar in practice is not executive habit but judicial requirement, and a declaration allowed to run without genuine review sits uneasily with the judgment. Consultation: the Central Government may make a declaration on its own motion, but it should ordinarily consult the state government before doing so, a softening of the consent problem that stops deliberately short of a veto. Force discipline: the power under Section 4 is to be exercised with the minimum force necessary for effective action, a standard the Court read into the section’s structure. And procedural discipline: the Court took the Army’s own list of do’s and don’ts for personnel operating under the Act and declared the instructions binding, including the requirements of cooperation with the civil administration, the handover of arrested persons to the nearest police station with the least possible delay, and the production discipline of the ordinary criminal law; it further held that a decision of the Central Government granting or refusing sanction under Section 6 must be supported by reasons, since the order is open to judicial review. The judgment is therefore best read as a bargain. The Act survived, but it survived as a conditioned Act, and the distance between the conditions as declared and the conditions as observed is the space in which the next three decades of litigation, commission reports, and protest were conducted. A critic of the judgment, and there have been many among scholars of the region, observes that conditions without an enforcement mechanism are closer to advice than to law: the six month review happens inside the same executive that benefits from the declaration, the consultation duty has no stated consequence if ignored, and the minimum force standard is tested only after a death, in proceedings that run into the Section 6 gate. A defender answers that the Court in 1997 was not writing on a blank page but sustaining a statute in the middle of live insurgencies, and that the conditions it declared have in fact been the hooks on which every later accountability proceeding, including the far more consequential judgment of 2016, has hung.

The petitioners’ arguments deserve a fuller statement than they usually receive, because the Court’s answers define the perimeter within which all later argument moves. On competence, the challenge pressed the point that the Act’s real subject was the maintenance of public order within states, and that dressing a public order law in the vocabulary of military deployment could not transfer a State List subject to Parliament. The Court answered with the doctrine of pith and substance: examined as a whole, the Act provides for the deployment of the armed forces of the Union in aid of the civil power, the forces do not supplant the civil administration, the state’s own government and police continue to function, and the incidental touching of public order does not change the law’s true character. On the emergency argument, the petitioners contended that the Act achieved, without a proclamation under Article 352, the substance of an emergency regime in the notified areas, suspending the ordinary discipline of criminal process while escaping the parliamentary approvals and time limits that attend a proclamation. The Court answered that the comparison fails at the decisive point: a disturbed area declaration suspends no fundamental right, displaces no state government, and confers no legislative power on the executive, and the rights of citizens under Articles 14, 19, and 21 continue to run against action taken under the Act, which is why the Act’s provisions could themselves be tested against those articles, and were. On arbitrariness, the attack on Section 4 fastened on the breadth of the firing power and the low rank of the officers holding it; the Court sustained the section by reading it with the conditions the section itself contains, the requirement of a prohibitory order, the necessity of the officer’s opinion for the maintenance of public order, the giving of due warning, and with the binding force of the do’s and don’ts and the handover discipline of Section 5. The sanction provision survived on the reasoning that previous sanction for acts done in the exercise of official power is a known and constitutionally accepted safeguard, provided the sanctioning authority applies its mind and records reasons that a court can review. Read carefully, the judgment is less a blank cheque than a structured settlement: each power survives because a discipline is attached to it, and the subsequent history of the Act is largely the history of whether those disciplines (review, consultation, minimum force, reasoned sanction) operate as the judgment supposed. That is also why the judgment, although it dismissed the challenge, is quoted by the Act’s critics no less than by its defenders: it supplies the standard by which the Act’s administration can be found wanting without any need to reopen the Act’s validity.

2016 and after: the end of absolute immunity

If 1997 defined the conditions, 2016 supplied the enforcement logic. Extra Judicial Execution Victim Families Association v Union of India arose from Manipur, where an association of families of persons killed in alleged fake encounters placed before the Supreme Court a documented list of 1,528 deaths at the hands of the police and the armed forces across the decades of the state’s insurgency, and complained that these deaths had met with no effective investigation at all. A commission appointed by the Court and headed by Justice Santosh Hegde examined a sample of the cases and reported in 2013 that the encounters it examined did not bear out the official accounts, a finding that hung over the proceedings when the Court gave judgment on 8 July 2016. The holdings that matter are three. First, the Court held that a declaration of a disturbed area does not place the armed forces beyond the law: the use of excessive or retaliatory force is not a protected exercise of AFSPA power, it does not matter whether the person killed was a militant or a civilian for the purpose of the duty to inquire, and every allegation of a death caused by excessive force in a disturbed area requires a thorough inquiry. Second, the Court held that there is no concept of absolute immunity for armed forces personnel: Section 6 regulates the institution of proceedings through the sanction requirement, but it does not convert an unjustified killing into a lawful act, and where an inquiry establishes excessive force, prosecution can follow through the ordinary criminal courts. Third, the Court directed investigation of the documented cases, a direction it converted in follow up proceedings into a court monitored investigation by a special investigation team of the Central Bureau of Investigation into the cases that had been documented through judicial inquiries and commission findings. The series thesis of this series of articles finds here its most consequential illustration: the operative content of a statute was transformed, the immunity that public debate treated as the Act’s core feature was declared not to exist in absolute form, and not one word of the statute’s text was amended by Parliament. The text of Section 6 in a bare Act volume reads exactly as it read in 1958; the law of Section 6 is found in the judgments.

The distinction the 2016 judgment draws, between the right to an inquiry and the path to a prosecution, is the distinction on which the modern accountability debate turns, and it must be stated exactly. Investigation is not gated by Section 6: a death in a disturbed area can and must be investigated, by the police, by a magistrate, by a commission, without anyone’s sanction. Prosecution is gated: the investigated file reaches a court only if the Central Government grants previous sanction. The practical consequence is a chain with one strong link and one weak link. The inquiry link has been strengthened by the courts to the point where an uninvestigated killing in a disturbed area is now plainly unlawful administration. The prosecution link remains an executive decision, taken by the government whose forces are accused, reasoned on paper after 1997 but unaccompanied by any statutory time limit or any deemed grant. Anyone assessing the claim that AFSPA confers total immunity should therefore be precise about dates and about stages. As a statement of law after 2016, the claim is wrong, and this article says so without qualification. As a statement about outcomes at the prosecution stage, the claim points at something real, and the evidence for it is the sanction record itself.

One passage of the 2016 judgment reaches beyond Manipur and speaks to the whole design, and it is the passage state governments quote most often. Confronted with a disturbed area regime that had operated in Manipur across decades, the Court treated longevity itself as a legal embarrassment: a situation of internal disturbance is by definition a departure from normalcy, and where the departure has continued for a generation, the continuation is evidence not of the regime’s necessity but of the failure of the ordinary machinery of governance, civil and military alike, to restore the normalcy the regime exists to protect. The observation does not invalidate any declaration, and the Court was careful to keep it at the level of principle. But it reverses the rhetorical default that longevity otherwise creates. Under the renewal machine, each six month extension borrows authority from the last, and a declaration renewed eighty times acquires the false solidity of a permanent institution. After 2016, the same arithmetic runs the other way in legal argument: every additional renewal of a decades old declaration is a further admission that the exceptional has failed to end the exception, and a government defending continuation must answer for the failure as well as for the danger. That is the doctrinal seed from which a future, more interventionist review of declarations would grow, and governments and their law officers are aware of it, which is one reason the reductions of notified areas in the years after the judgment were presented publicly as achievements of policy rather than as the administrative routine the renewal machine usually performs in silence.

The accountability chain and its weakest link

The sanction record is difficult to summarise in figures, because no official consolidated account of Section 6 decisions across all states and decades is published, and any article that offers a precise global percentage is inventing it. What the record does contain is a series of disclosed episodes, in legislative answers, in affidavits, and in litigated cases, and the episodes point in one direction: requests are few relative to the number of allegations, decisions are slow, and refusals are the norm in the cases that attract public attention. The episode that completes the picture is the most fully documented one. After the killings in Mon district in December 2021, the Nagaland Police’s special investigation team investigated the operation and filed a chargesheet against thirty personnel of the unit involved. Prosecution required the Centre’s sanction under Section 6. The competent authority declined sanction by an order dated 28 February 2023. The Supreme Court, which had stayed the proceedings in 2022 for want of sanction, thereafter closed the criminal proceedings altogether, holding that the bar in Section 6 left the courts no room to proceed while sanction stood refused, while recording that the proceedings could revive if sanction were granted in the future and that internal disciplinary action under Army law remained open. The State of Nagaland, in the extraordinary position of a state whose own police had chargesheeted soldiers its own assembly wanted tried, challenged the refusal of sanction by a separate petition. The episode is the consent problem and the accountability problem in a single file: the state investigated, the state prosecuted on paper, the state demanded, and the decisive signature belonged throughout to the Centre.

The same structural pattern repeats in the human rights machinery. The National Human Rights Commission, which for complaints against other public servants can inquire on its own authority, meets a statutory wall where the armed forces are concerned. Under the Protection of Human Rights Act, 1993, whose Section 19 prescribes a special procedure for the armed forces, the Commission may not investigate a complaint against the forces itself: it may seek a report from the Central Government, and on receiving it may either drop the complaint or make recommendations to that government, which must report the action taken within three months. The design means that for the category of allegation most likely to arise under AFSPA, the national human rights institution is a recipient of the accused institution’s own government’s report, a limitation that the Commission itself has repeatedly criticised and that reform proposals have repeatedly targeted. The full working of the Commission’s powers and their limits belongs to the institution’s own articles in this series; what matters here is the consequence. Between a sanction gate held by the Centre for prosecutions and a report based procedure for the human rights commission, every formal accountability channel for an act done under AFSPA runs, at the decisive point, through the Central Government. That is not an accident of administration. It is the statute book working as drafted, and it is why the federalism framing of this article is not a decorative alternative to the human rights framing but a description of the same machinery from the other end. The follow through of the 2016 judgment shows both the reach and the limit of the judicial method. The Court kept the Manipur matter before it, required the petitioners and the authorities to complete the documentation of the alleged killings through the records of judicial inquiries, commission findings, and human rights proceedings, and in 2017 directed the Central Bureau of Investigation to constitute a special investigation team and investigate the documented cases, beginning with a tranche of ninety eight cases that inquiries had already examined. The direction converted the abstract duty to investigate into named files and a reporting obligation to the Court. Yet the endpoint of even a completed investigation remains the sanction gate, and the Manipur process, like the Mon process after it, illustrates the asymmetry the judgment could not by itself remove: courts can command that a death be investigated, can monitor the investigation, and can condemn an unreasoned refusal of sanction, but the decision to let a prosecution begin is assigned by the statute to the executive, and an executive that declines, with reasons recorded, leaves the courts to police the reasoning rather than supply the decision. The sanction decision’s internal journey reinforces the point. A request travels from the investigating police through the state government to the Centre, where for Army personnel the decision rests with the Central Government advised through the defence establishment whose personnel stand accused, a structure that no one designed for AFSPA specifically but that the Act’s centralising draftsmanship reproduces at the accountability stage exactly as at the declaration stage.

The surrounding inquiry framework sharpens the asymmetry by contrast. For deaths at the hands of the police, Indian law and practice supply comparatively dense machinery: the National Human Rights Commission’s guidelines on encounter deaths require prompt reporting of every such death to the Commission and inquiry through a magistrate, and the criminal procedure requires a judicial magistrate’s inquiry into deaths in custody. For deaths at the hands of the armed forces in a disturbed area, the machinery thins at exactly the institutional points this article has identified: the Commission’s own inquiry power is displaced by the report procedure of Section 19 of the human rights statute, and prosecution awaits a sanction that investigation cannot compel. The result is a two track accountability system whose tracks correspond to the identity of the force involved rather than to the gravity of the allegation, and reform proposals that focus on the tracks, mandatory judicial inquiry into every death in a disturbed area, a Commission freed of the Section 19 restriction for such cases, a published sanction record, are aimed at closing a gap that the judgments have exposed but cannot themselves legislate away. A researcher assembling the record of any single contested death will move between these tracks constantly, from police first information reports to magisterial findings to commission reports to sanction correspondence, and the scatter of that record across institutions is itself a finding about how the system is built.

Behind the 2016 judgment stood a documentary achievement that deserves naming, because it changed what could be proved. The list of 1,528 deaths was not a rhetorical figure: it was a compiled record, assembled over years by Manipuri civil society organisations and human rights documentation groups from first information reports, newspaper accounts, commission records, and the testimony of families, organised case by case so that each entry could be tested. The Supreme Court’s procedure in the case honoured that method: rather than pronounce on the aggregate, it directed the parties to sort the cases by the state of their documentation, required the production of the records of judicial inquiries and commission proceedings that had touched individual deaths, and built its investigation directions on the documented residue. The lesson generalises across the Act’s theatres. In a legal regime where the operative facts are dispersed among police stations, military units, and gazette files, the party that documents wins the right to be heard, and the long, unglamorous labour of compiling names, dates, and records by local organisations is the foundation on which the constitutional litigation of 2016, and the political pressure that preceded and followed it, ultimately rested. The High Courts of the region carry the corresponding daily docket: habeas corpus petitions during the handover interval, compensation claims after contested deaths, and challenges to sanction refusals arrive at the Gauhati High Court and the Manipur and other regional High Courts in a steady stream, and their orders, less celebrated than the two Supreme Court judgments, are where the conditions of 1997 are most often enforced in individual cases. An account of AFSPA’s judicial life that stopped at the Supreme Court would miss the level at which the law is most frequently applied to the state, one petition at a time.

Two practical notes belong in this section. The documentary trail of any AFSPA controversy is unusually scattered, spread across gazette notifications, inquiry reports, chargesheets, and sanction orders that circulate as scanned files and tabulated lists of uneven quality; a researcher assembling that trail can open such material privately in the browser using the ReportMedic office file viewer for opening case files, inquiry records and tabulated documentation, which renders spreadsheets, documents, and presentations on the reader’s own device without uploading them. And nothing in this account of the sanction process is advice on any individual case: a family pursuing a complaint faces limitation periods and forum rules that turn on the facts of the matter, and the general position described here does not substitute for advice on a specific case, which should be taken before any step is filed.

The case for the Act, stated in its strongest form

The retention case begins from a proposition that repeal advocates sometimes skip: the alternative to AFSPA is not the absence of military deployment but military deployment without a statutory frame. Where an insurgency exceeds police capacity, governments will deploy the Army in aid of civil power under executive authority, Emergency provisions, or improvised legal cover, and the soldier on the ground will act either way. The Army’s institutional argument, made consistently in official evidence, in ministerial statements, and in the Union’s affidavits in the constitutional litigation, is that a soldier so deployed needs three things that only a statute can supply: a defined legal basis for the use of force beyond the narrow defences available to any citizen, protection against prosecution for acts done in good faith in an environment where split second decisions are reviewed years later in safety, and clarity about the chain of authority under which he acts. On this view Section 4 is not an indulgence but a specification: it tells the soldier what he may do, after what warning, against whom, and Section 6 is not impunity but the standard Indian device, found across the statute book for public servants, of filtering vexatious or retaliatory prosecution through a sanction screen. The strategic limb of the argument adds that AFSPA theatres are border theatres: the north-eastern states and Jammu and Kashmir face armed groups sustained from across international borders, the counter-insurgency grid of posts and patrols is what keeps those groups from consolidating territory, and withdrawing the legal frame mid conflict signals to armed groups that endurance is a winning strategy. The empirical limb points to the map’s contraction: Punjab’s insurgency ended, Tripura’s ended, Mizoram settled, and in each case the exceptional law left with the violence, which is offered as proof that the Act is an instrument governments use against insurgency rather than a device for permanent control. The strongest version of the retention case finally turns the accountability argument around: the remedy for abuse of a necessary power is the enforcement of the conditions the courts have already written, investigation, sanction decisions taken honestly and with reasons, and prosecution where force was excessive, and the existence of cases where that chain failed is an indictment of the chain’s operators rather than of the statute that also contains it.

The Army’s own doctrinal writing belongs in a fair statement of its case. The force’s published doctrine for sub conventional operations presents counter-insurgency as a population centric undertaking in which force is to be calibrated, minimum force is the governing standard, and the winning of the population’s confidence is an operational objective rather than a public relations gloss, a philosophy its officers summarise in the formula of an iron fist inside a velvet glove. The significance of the doctrine for the legal debate is that the standards by which the courts have judged the Act, minimum force, restraint after arrest, cooperation with civil authority, are standards the Army professes for itself, so that enforcement of the 1997 conditions can be presented, and often is presented by serving officers, not as an external imposition but as the legal form of the Army’s own professional ethic. The gap that critics identify is therefore not between the Army’s doctrine and the law’s demands but between both and the record of particular operations, and the institutional question is why a force whose doctrine and a court whose conditions agree so closely have jointly failed to discipline the exceptions.

Each limb of that case deserves the respect of a direct answer. The defined basis argument is genuinely strong: no serious participant in the debate proposes that soldiers in a firefight should rely on the private defence provisions of the penal code, and the Jeevan Reddy Committee, the very emblem of the repeal case, accepted the need for a legal frame by proposing to rebuild one inside another statute. The border argument is a real strategic consideration in theatres where armed groups operate from foreign soil. The weak points of the retention case lie elsewhere. The contraction argument proves that AFSPA ends when insurgency ends, which no one disputes, but it does not answer the complaint that in the longest theatres the six month review has operated as renewal by default, with the burden of proof effectively on those seeking withdrawal rather than on those seeking to continue an exceptional regime into its fifth and sixth decades. The sanction screen argument depends on the screen being operated as a filter rather than as a wall, and the documented record, culminating in a chargesheeted case closed without trial after the Mon killings, is the retention side’s hardest exhibit to explain. And the clarity argument cuts both ways: a statute that gives clarity to the soldier and leaves the citizen to discover the operative conditions in three decades of case law has distributed its clarity asymmetrically. The retention case, fairly stated, establishes that some statute is necessary where the Army is deployed against insurgency. It does not by itself establish that this statute, with this declaration mechanism and this sanction record, is the statute that necessity requires.

The case for repeal, and the committees that made it

The repeal case has the unusual distinction of having been made, twice, by the state’s own appointees. In November 2004, after a summer of protest in Manipur over a death in the custody of the Assam Rifles and the withdrawal of the Act from the Imphal municipal area, the Central Government appointed a five member committee under Justice B. P. Jeevan Reddy, a former judge of the Supreme Court, to review the Act. The committee reported on 6 June 2005, and its central recommendation was unqualified: the Act should be repealed. The committee found that the statute had become, in its words, a symbol of oppression, an object of hate and an instrument of discrimination and high handedness, and it proposed that the provisions genuinely needed for counter-insurgency be absorbed into the Unlawful Activities (Prevention) Act, accompanied by safeguards and by a grievance redressal mechanism, so that the powers would operate inside a general security statute rather than under a region specific regime carrying six decades of accumulated symbolism. The government neither accepted nor formally rejected the report for years; it was not officially published, and it entered public debate through unofficial circulation. Two years later the Second Administrative Reforms Commission, chaired by Veerappa Moily, in its report on public order of 2007, endorsed repeal in similar terms, describing the Act as too sketchy and drastic for the work it does and recommending its replacement by a more carefully drawn framework. In 2013 the committee under Justice J. S. Verma, appointed after the Delhi gang rape to review laws on sexual violence, addressed the Act from a different angle and recommended that sexual offences by armed forces personnel in conflict areas be tried under the ordinary criminal law rather than filtered through the protective machinery, and that the operation of the Act be reviewed. Three official exercises, across a decade, under governments of different composition, converged on the same conclusion by different routes: the statute in its enacted form could not be defended as drafted.

The popular repeal case adds two elements the committees only implied. The first is duration as an argument in itself. A law designed for a defined emergency in 1958 has operated in parts of Nagaland continuously for over six decades, in Manipur for over four, through generations of citizens who have lived their entire lives under a legal regime that the Constitution treats as exceptional; the critic’s phrase for this, a permanent state of exception, names the intuition that a measure’s legitimacy is partly a function of its temporariness, and that renewal by six month increments has dissolved the distinction between an emergency instrument and a system of government. The second element is the testimony of protest, of which the defining instance is Irom Sharmila’s fast in Manipur: begun in November 2000, days after ten civilians were shot dead at a bus stop in Malom by personnel of the Assam Rifles, and sustained for sixteen years until August 2016, through repeated arrest and force feeding under the law against attempted suicide, it made a single citizen’s body into a standing referendum on the Act in Manipur, and its length became itself an argument about the regime it opposed. Governments of both national parties received these recommendations and this protest and declined to act, for the reasons the retention case supplies: the Army’s institutional opposition, the fear of a security vacuum in live theatres, and the political calculation that the costs of repeal are concentrated and immediate while the costs of retention are borne by distant populations. The Jeevan Reddy report’s afterlife poses the sharpest question in the whole debate. A state that appoints its own reviewers, receives a unanimous direction from them twice over, and shelves the result without a parliamentary debate on the merits has not answered the repeal case; it has declined to have it, and the distinction between those two things is visible to every citizen of the affected states.

Parliament’s part: questions asked, answers held

For a statute whose continuation is an executive routine, Parliament’s role has been confined to interrogation, and the record of that interrogation is remarkably static. Across decades, members from the affected states and from the opposition benches have raised the Act through questions, calling attention motions, and debates on the north-east and on Kashmir, pressing the demands that recur throughout this article: publication of the Jeevan Reddy report, a statement of the sanction record, a timetable for withdrawal, the reasons for each renewal in states whose governments object. The executive answer, delivered by the Ministry of Home Affairs under governments of both national parties, has held a consistent shape: the Act is reviewed periodically, the notified areas are reduced as the security situation improves, the armed forces require the legal protection the Act provides, and decisions on sanction are taken on the merits of each case. Private initiatives to repeal or amend the Act have been introduced from time to time and have not been taken up for passage. The significance of the record is structural rather than partisan. A legislature that enacted the original measure in a debated vote in 1958, and extended it in 1972, has since participated in the regime only as a questioner, because the regime’s operative decisions, declaration, renewal, sanction, are all executive acts that never come before it. When critics describe AFSPA as a failure of democracy, the precise content of the charge is this distribution of roles: the most consequential internal security regime in the Republic operates, decade after decade, without any legislature being asked to approve its continuation in any place at any time. The reform proposals that would restore a legislative moment, parliamentary ratification of long lived declarations, a statutory sunset requiring reauthorisation, are aimed at exactly this absence, and their repeated non adoption is the measure of how completely the executive routine has displaced the legislative one.

December 2021: Mon, the resolutions, and the limits of a state’s voice

The federal structure of the Act was displayed with unusual completeness after the events of December 2021 in Nagaland’s Mon district. On 4 December 2021, personnel of an Army special forces unit, acting on intelligence about the movement of an armed group, fired on a vehicle carrying coal mine workers returning to Oting village, killing six of them; in the violence that followed as villagers confronted the unit, further firing killed seven more civilians, and a soldier also died. The next day, in Mon town, security forces fired on an agitated crowd and one more civilian was killed, bringing the civilian dead to fourteen, a sequence of events stated here without embellishment because the bare sequence is the material fact. The political response inside Nagaland was total and cross party. The state government constituted a special investigation team of the Nagaland Police, which investigated and chargesheeted thirty personnel. On 20 December 2021 the Nagaland Legislative Assembly, in a special session, unanimously resolved to demand that the Government of India repeal the Act from the north-east and specifically from Nagaland, and the chief ministers of Nagaland and Meghalaya publicly demanded repeal, a demand echoed across the region’s political spectrum. A state government had done everything a state government can do: it had investigated with its own police, prosecuted on paper, and spoken with a single voice through its legislature. The sequel has been described in the accountability section of this article: sanction to prosecute was refused in February 2023, and the courts closed the proceedings the state’s police had built. The Mon sequence should be taught whenever the Act is taught, because it removes every ambiguity from the federalism analysis. No one in December 2021 disputed the state’s sincerity, its competence to judge conditions in Mon district, or its constitutional responsibility for public order within Nagaland. The Act operated anyway, on central notifications, and its accountability machinery answered to the Centre. A reader who understands Mon understands Section 3; a reader who has only debated Section 4 does not yet understand the statute.

The investigation of Mon proceeded on two tracks from the first week, and the duality is characteristic of accountability under the Act. The state track was the Nagaland Police investigation described earlier, conducted by a special investigation team answerable to the state government, which visited the site, recorded civilian testimony, and chargesheeted thirty personnel. The military track was the Army’s own Court of Inquiry, an internal fact finding proceeding under military law, convened to establish what the unit had done and why, whose findings feed the force’s disciplinary system rather than any public record. The two tracks answer to different masters, apply different instruments, and publish on different principles: the police chargesheet became a public document through the courts, while the Court of Inquiry’s findings remained within the Army, summarised for the public only in official statements. A family seeking the whole truth of a single night must therefore pursue it in fragments, a chargesheet here, a sanction file there, an internal inquiry whose contents it cannot compel, which is a further reason the documentary labour of civil society, and the courts’ insistence that investigation cannot be gated even where prosecution can, carry the weight they do in this system.

AFSPA and the peace processes: the statute and the settlements

The Act’s relationship with the region’s peace processes is double sided, and both sides are argued by participants rather than commentators. On one side stands the security establishment’s sequence: the Act creates the conditions of order within which negotiation becomes possible, armed groups come to the table when the military balance leaves them no better option, and the settlements that ended the Mizoram insurgency in 1986, subdued Tripura’s, and produced the ceasefire with the principal Naga armed group in 1997 were all reached while the Act stood in force over the negotiating parties. On the other side stands the experience of societies negotiating under a disturbed area regime: the Naga talks have proceeded for more than two decades under a ceasefire whose ground rules regulate the movement of armed cadres, while the state in which the talks occur remains notified as disturbed and the Act’s powers remain available against the very population whose political future is under negotiation, a juxtaposition that Naga political society cites as evidence that the Union negotiates with one hand and reserves coercion with the other. The Nagaland Assembly’s resolution of December 2021 made the connection explicit, demanding repeal in order to strengthen the efforts toward a peaceful political settlement of the Naga issue. The Framework Agreement of 2015 between the Union and the Naga leadership raised expectations of a final settlement, and the continuing presence of the Act in Nagaland thereafter became, for the state’s political class across parties, the standing symbol of the settlement’s incompleteness. The analytical point survives the rival slogans. A disturbed area declaration is a finding that a territory is too dangerous for ordinary law; a peace process is a claim that the same territory is ready for political normalcy; and a state cannot indefinitely assert both about the same districts without one of the assertions losing credit. The withdrawals that have held, Mizoram after 1986 and Tripura after 2015, were sequenced as part of settlements or their consolidation, and the sequence is the strongest practical evidence in the whole debate that the Act’s ending, when it comes, arrives as a political act attached to a political settlement, and not as the expiry of a legal instrument or the conclusion of a court case.

The reform options actually on the table, and a verdict

Reform discussion of AFSPA tends to be reported as a binary, repeal or retain, but the working menu is longer, and most items on it are Section 3 reforms rather than Section 4 reforms, which tells its own story about where the statute’s problems are located. The Jeevan Reddy route is repeal with transplantation: abolish the Act and carry the necessary powers into the general anti terrorism statute with added safeguards, accepting the Army’s need for a legal frame while ending the region specific exceptional regime. A second route amends the declaration mechanism itself: require the concurrence of the state government for a central declaration, or require parliamentary ratification of declarations beyond a stated period, or write a genuine sunset into Section 3 so that a declaration lapses unless affirmatively justified afresh against published criteria. A third route amends Section 6: impose a time limit on sanction decisions with a deemed grant on expiry, require the reasons the 1997 judgment demanded to be published, or transfer the sanction decision for the gravest offences to an authority outside the accused force’s own government. A fourth route strengthens the surrounding machinery rather than the Act: free the National Human Rights Commission from the Section 19 report procedure for armed forces complaints, mandate time bound judicial inquiry into every death in a disturbed area, and publish a consolidated sanction record so that the debate can proceed from figures rather than episodes. Each route has a constituency and a history of official endorsement; none has been taken, and the six month renewal cycle has outlasted every committee appointed to interrupt it.

Rebuilding Section 3: what an amendment would actually have to say

Reform proposals acquire force when they are stated at the level of the text, because the Act’s defects are textual and the drafting choices are identifiable one by one. A consent based Section 3 would provide that a declaration by the Central Government over the territory of a state shall not be made except with the concurrence of the government of that state, or, in the softer version canvassed in federal debate, that such a declaration shall cease to operate at the end of a stated period unless the state government concurs in its continuance, a drafting device that reverses the burden of persistence without denying the Centre the opening move in a genuine crisis. A ratification based Section 3 would leave the declaration power untouched and instead provide that a declaration continued beyond a cumulative period, two years being the figure most often proposed in debate, shall not remain in force unless approved by resolution of Parliament, converting the renewal machine’s output into a periodic legislative act and restoring to the legislature the role the extension history removed from it. A sunset based Section 3 would attack duration directly, providing that no area shall remain continuously declared for longer than a stated maximum except under a fresh enactment addressed to that area, the strongest and least likely of the three, since it requires Parliament to legislate theatre by theatre as it did in 1958, 1983, and 1990. On the accountability side, a rebuilt Section 6 would replace the open ended sanction power with a time bound one: the Central Government to decide within a stated period, failing which sanction is deemed to have been granted, with reasons recorded and published in either event, the deemed grant being the standard drafting answer wherever sanction regimes have been reformed, because it abolishes the pocket veto that the present text permits. None of these devices is exotic. Consent clauses, ratification requirements, sunsets, and deemed sanctions all appear elsewhere in the Indian statute book, and the Sarkaria framework supplies the federal theory for the first of them. Their collective absence from AFSPA is therefore not a drafting tradition but a drafting decision, renewed by every government that has declined to amend, and stating the alternatives textually has the polemical advantage of forcing the retention case to defend the specific defaults, central signature, silent renewal, timeless sanction, rather than the general necessity of a legal frame, which the reform case concedes.

The verdict this article defends follows from its namable claim. The question that decides AFSPA’s legitimacy is not whether soldiers in a counter-insurgency need statutory powers and statutory protection; they do, and the repeal case at its best has always conceded it by proposing a replacement frame. The deciding question is who should hold the power to place and keep a part of a state under an exceptional regime, and on that question the Act’s design is indefensible in a federation whose Constitution assigns public order to the states: a declaration the Centre can issue without concurrence, renew without fresh justification to any legislature, and pair with a prosecution gate it controls itself, concentrates every decision that matters in the hands of the party that also conducts the operations. The courts repaired what they could reach, striking the absolute immunity out of the law in 2016 and writing review and consultation disciplines into the declaration power in 1997, and the gap between the Act as the courts have conditioned it and the Act as it is administered is now the most honest measure of the problem. Two Supreme Court decisions changed what this statute means without Parliament changing a word of it. Whether Parliament eventually does the work itself, by repeal and transplantation or by rebuilding Section 3 around consent, sunset, and a sanction process that can survive scrutiny, is the unfinished business the next decade of this statute’s long life will either settle or repeat.

Frequently Asked Questions

Q: What is AFSPA and where does it apply?

AFSPA is the Armed Forces (Special Powers) Act, 1958, a central statute that gives the armed forces of the Union special powers, including the power to use force after warning, to arrest without warrant, and to search without warrant, inside areas declared disturbed under its Section 3. It was enacted for Assam and Manipur and extended across the north-eastern states, where it operates across whole states or in notified parts of states depending on the security assessment recorded in the live notifications. Jammu and Kashmir is governed by a separate Act of 1990 modelled on it. The Act applies only where a declaration is in force, so its operative map is the set of live notifications, which changes as areas are added or removed.

Q: Who declares an area disturbed under AFSPA?

A disturbed area declaration under Section 3 may be made by three authorities acting independently: the Governor of the state, the Administrator of a union territory, or the Central Government, the last of these having received the power by the amendment of 1972. The declaration is made by notification in the Official Gazette and records the authority’s opinion that the area is in such a disturbed or dangerous condition that the use of the armed forces in aid of the civil power is necessary. No declaration requires the concurrence of any other authority, which is why a central declaration can operate in a state whose government opposes it, and why state demands for withdrawal are addressed to the Centre as requests rather than exercised as rights.

Q: Can a state government get AFSPA withdrawn on its own?

Only in part. Where the operative notification is the Governor’s, a state government can ordinarily secure withdrawal through the Governor acting on its advice, as the government of Tripura did in 2015. Where the operative notification is the Central Government’s, the state has no power of revocation: it can pass assembly resolutions, send formal recommendations, and press its case with the Ministry of Home Affairs, but the notification continues until the Centre withdraws it or allows it to lapse. The Nagaland Assembly’s unanimous resolution of December 2021 demanding repeal illustrates the limit exactly. The state’s voice was complete and united, and the Act continued in force because the decisive notification power lay elsewhere.

Q: Does AFSPA give the armed forces total immunity from prosecution?

No, not as a matter of law since the judgment of 2016. Section 6 of the Act bars any prosecution, suit, or other legal proceeding for acts done under the Act except with the previous sanction of the Central Government, and for decades that gate made the protection close to complete in practice. In Extra Judicial Execution Victim Families Association v Union of India (2016), the Supreme Court held that there is no absolute immunity under the Act: the use of excessive or retaliatory force is not protected, every such death must be investigated, and prosecution may follow where an inquiry establishes unjustified force. Sanction is still required before a prosecution begins, and it is still rarely granted, so the practical position remains closer to the criticism than the legal position suggests.

Q: Did the Supreme Court uphold AFSPA?

Yes. In Naga People’s Movement of Human Rights v Union of India, decided on 27 November 1997, a five judge Constitution Bench upheld the Act’s constitutionality against the arguments that Parliament lacked competence over public order and that the Act created an emergency without a proclamation. The Court held the Act to be a law on deploying the armed forces in aid of civil power. The upholding came with conditions the Court read into the statute: declarations must be for a limited duration with periodic review before six months expire, the Centre should ordinarily consult the state before declaring on its own motion, the force used must be the minimum necessary, and the Army’s own list of do’s and don’ts is binding on personnel operating under the Act.

Q: What did the Jeevan Reddy Committee recommend about AFSPA?

The committee headed by Justice B. P. Jeevan Reddy, appointed in 2004 to review the Act and reporting on 6 June 2005, recommended that AFSPA be repealed outright. It described the Act as a symbol of oppression and an instrument of discrimination and high handedness, and proposed that the provisions genuinely needed for counter-insurgency be incorporated into the Unlawful Activities (Prevention) Act with safeguards and a grievance redressal mechanism, so that the powers would operate under a general security statute rather than a region specific regime. The recommendation was never implemented: the report was not officially published for years, governments neither accepted it nor debated it in Parliament, and the Second Administrative Reforms Commission’s endorsement of repeal in 2007 met the same fate.

Q: From which parts of India has AFSPA been lifted over time?

The Act has been fully withdrawn from some states and reduced in others. Punjab’s companion statute of 1983 was repealed in 1997 as its insurgency ended, and Mizoram passed out of the Act’s geography after the peace accord of 1986. The government of Tripura withdrew the notification in May 2015 after eighteen years, and the Centre removed the notification from Meghalaya’s border belt with effect from 1 April 2018. Manipur’s capital was taken partly out in 2004 when the Imphal municipal area lost its disturbed status, and in 2022 substantial sets of districts and police station areas in Assam, Manipur, and Nagaland were removed from the notifications, while the retained areas continued on six month renewal cycles.

Q: Is AFSPA a law and order subject belonging to the states?

The subject matter points both ways, which is why the question was litigated. Public order and police are Entries 1 and 2 of the State List, so the states begin with the stronger textual claim. Parliament’s power rests on Entry 2A of the Union List, covering deployment of the Union’s armed forces in aid of the civil power, reinforced by Article 355, which obliges the Union to protect every state against internal disturbance. The Supreme Court held in 1997 that AFSPA is in substance a law on deploying the armed forces in aid of civil power rather than a law on public order, and therefore within Parliament’s competence. The legal question is settled; the federal grievance that a state subject is being administered from Delhi survives the settlement.

Q: Why was AFSPA enacted in 1958?

AFSPA was enacted to give the Indian Army a statutory basis for operations against the Naga insurgency in the Naga Hills, then a district of Assam. The Assam government’s own instruments, including the Assam Disturbed Areas Act of 1955, had armed the police and paramilitary forces, but the campaign against the Naga National Council’s parallel administration was judged to require the Army, and the Army required legal cover for using force, arresting, and searching in a civilian theatre. The President first promulgated an ordinance on 22 May 1958, and the Armed Forces (Assam and Manipur) Special Powers Act followed on 11 September 1958, modelled in structure on the colonial Armed Forces (Special Powers) Ordinance of 1942.

Q: What test must be satisfied before an area is declared disturbed under AFSPA?

The test is the one Section 3 states, and it is a test of opinion rather than of measured fact. The Governor of the state, the Administrator of the union territory, or the Central Government must be of the opinion that the whole or the part in question is in such a disturbed or dangerous condition that the use of the armed forces in aid of the civil power is necessary, and must publish the declaration by notification in the Official Gazette. The Act fixes no threshold of violence, no casualty figure, and no species of disorder that qualifies, so an area is disturbed in law once the opinion is formed and notified, and it stops being disturbed when the notification is withdrawn or allowed to lapse.

Q: Is Jammu and Kashmir covered by the 1958 Act or by a separate law?

Jammu and Kashmir is covered by a separate statute, the Armed Forces (Jammu and Kashmir) Special Powers Act, 1990, and not by the Act of 1958. The 1990 Act is closely modelled on the earlier statute: it carries its own disturbed area declaration mechanism, its own grant of special powers to the armed forces, and its own requirement of previous sanction before prosecution. The distinction is not pedantry. Notifications, sanction decisions, and litigation concerning Kashmir belong to the record of the 1990 Act, and accounts that attribute the history of the 1958 Act to Kashmir, or the operation of the 1990 Act to the north-east, misstate which law is doing the work in each theatre.

Q: Did any official committee other than the Jeevan Reddy Committee recommend repealing AFSPA?

Yes. The Second Administrative Reforms Commission, chaired by Veerappa Moily, recommended the repeal of AFSPA in its report on public order of 2007, endorsing the direction of the Jeevan Reddy Committee and treating the Act as too sketchy and drastic for the work assigned to it, to be replaced by a more carefully drawn legal framework. In 2013 the committee headed by Justice J. S. Verma, appointed to review the laws on sexual violence, recommended that sexual offences by armed forces personnel in conflict areas be tried under the ordinary criminal law and that the working of the Act be reviewed. No government has acted on any of these recommendations, which is why the repeal case can be constructed entirely from official sources.

Q: What happened after the killings in Mon district in December 2021?

After the killings of fourteen civilians in Mon district on 4 and 5 December 2021, the Nagaland government appointed a special investigation team of the state police, which investigated the Army unit involved and filed a chargesheet against thirty personnel. The Nagaland Legislative Assembly, meeting in special session on 20 December 2021, unanimously resolved to demand the repeal of AFSPA from the north-east and specifically from Nagaland. Prosecution, however, required the Central Government’s sanction under Section 6 of the Act; the competent authority refused sanction on 28 February 2023, and the Supreme Court closed the criminal proceedings for want of sanction, leaving open their revival if sanction is ever granted, together with disciplinary proceedings under Army law.

Q: Why did Irom Sharmila fast against AFSPA for sixteen years?

Irom Sharmila, a Manipuri civil rights activist, began a hunger fast in November 2000, days after ten civilians were shot dead at a bus stop in Malom by personnel of the Assam Rifles, demanding the repeal of AFSPA from Manipur. She sustained the fast for sixteen years, until August 2016, spending much of that period in judicial custody and being fed through a nasal tube after repeated arrest on charges of attempting suicide. Her protest became the most widely recognised symbol of opposition to the Act in Manipur and carried the region’s grievance to national and international attention, although it did not by itself change the statute: the Act remained in force in Manipur when she ended the fast and turned to electoral politics.

Q: What are the do’s and don’ts the Army must follow under AFSPA?

The do’s and don’ts are the Army’s own list of instructions for personnel operating under AFSPA, which the Supreme Court declared legally binding in 1997 rather than merely advisory. Their substance includes acting only within a declared disturbed area, using the minimum force necessary for effective action, not using force against a person after arrest except to prevent escape, not subjecting arrested persons to third degree methods, not interrogating arrested persons instead of handing them over, handing every arrested person to the nearest police station with the least possible delay, rendering medical assistance to injured persons, keeping accurate records of operations, and cooperating with the civil administration rather than acting as an independent authority. A breach of the list is a disciplinary matter and bears on whether the force used in an operation was lawful.

Q: Can armed forces personnel face court martial even when prosecution sanction is refused?

Yes. Section 6 of AFSPA controls proceedings instituted in the ordinary courts; it does not displace the armed forces’ internal jurisdiction under the Army Act. Personnel remain subject to military law for conduct connected with their service, and the force may convene a court martial or take other disciplinary action on the same facts without any sanction from the Central Government, because no external prosecution is being instituted. The Supreme Court made the point expressly when it closed the Mon district proceedings in the absence of sanction, recording that its order did not affect the Army’s liberty to take disciplinary action. Families of victims often regard this route as unsatisfying, because its proceedings are internal and its outcomes are not published in the manner of a public trial.

Q: Is AFSPA a temporary law with an expiry date?

AFSPA has no expiry date, no sunset clause, and no provision requiring Parliament to reauthorise it. The Act of 1958 is a permanent statute that remains in force until Parliament repeals it. What expires and renews is not the Act but the disturbed area declarations made under it, which the Supreme Court held in 1997 must be of limited duration and reviewed before six months pass. The contrast is the point critics press: the statute in its text is permanent, while each application of it is supposed to be temporary and freshly justified. In the longest running theatres the declarations have been renewed in six month cycles for decades, which is how a permanent Act and an ostensibly temporary regime coexist.

Q: What did the Supreme Court decide in the Manipur extrajudicial killings case?

In Extra Judicial Execution Victim Families Association v Union of India, decided on 8 July 2016, the Supreme Court considered a documented list of 1,528 alleged extrajudicial killings in Manipur. The Court held that a disturbed area declaration does not place the armed forces beyond the law, that the use of excessive or retaliatory force is not protected by AFSPA, that it makes no difference to the duty of inquiry whether the dead person was a militant or a civilian, and that there is no absolute immunity for armed forces personnel: every such death requires a thorough investigation, and prosecution can follow where excessive force is established. The Court then directed a court monitored investigation into the documented cases.

Q: Why was AFSPA withdrawn from Tripura and Meghalaya?

In both cases the stated reason was the security assessment. Tripura’s insurgency, which had brought the Act to the state in 1997, declined over years of counter-insurgency, surrenders, and political negotiation to a level the state government judged manageable by the police under the ordinary law, and the government headed by Manik Sarkar withdrew the notification in May 2015 on that assessment. Meghalaya’s case was narrower: the Act had only ever operated in a belt of twenty kilometres along the Assam border, and the Centre withdrew the notification from that belt with effect from 1 April 2018 as militancy in the Garo Hills declined. Withdrawal in both places was an executive judgment about conditions on the ground, taken as violence fell, and not the result of litigation or of any change in the Act itself.

Q: Does AFSPA give its special powers to the state police as well as the armed forces?

No. The special powers in Section 4 are conferred on members of the armed forces, which the Act defines as the military forces and the air forces operating as land forces, together with any other armed forces of the Union so operating, a definition that covers the Army and central forces such as the Assam Rifles when deployed in aid of civil power. State police personnel receive no Section 4 powers from the Act, and a state police officer in a disturbed area works under the ordinary criminal law that governs policing in the rest of India. The distinction matters in Manipur, where many contested killings involved state police commandos rather than the Army and were investigated as ordinary offences, with no AFSPA sanction question arising at all.