ED Investigation Powers and Limits

The Enforcement Directorate (ED) wields a battery of coercive powers under the Prevention of Money Laundering Act, 2002 that are, in their statutory design, far wider than those of an ordinary police officer under the CrPC. It can summon and interrogate, search and seize, freeze and attach property before any conviction, and arrest on its own satisfaction. Yet every one of these powers carries an internal trigger and an external limit, and since Vijay Madanlal Choudhary v. Union of India (2022) the Supreme Court has spent three years recalibrating where the muscle ends and the safeguard begins. This chapter maps the ED's investigative arsenal section by section, then traces the case law that has turned abstract "safeguards" into enforceable rights.

Who Is the ED, and Where Do Its Powers Come From?

The Enforcement Directorate is the agency entrusted with enforcing the PMLA. Its officers are not constituted as "police" but as authorities under the Act, drawing their coercive powers directly from Chapter V ("Summons, Searches and Seizures, etc.") and the arrest power in Section 19. This distinction is not cosmetic. In Vijay Madanlal Choudhary v. Union of India (2022 INSC 757, decided 27 July 2022), a three-Judge Bench held that ED officers are not police officers within the meaning of the Code of Criminal Procedure, with the consequence that the bar in Section 25 of the Evidence Act on confessions to police does not automatically apply to statements recorded by the ED.

The same judgment confirmed that the offence of money laundering under Section 3 is a standalone, continuing offence built on "proceeds of crime" generated by a scheduled (predicate) offence. The ED's powers are therefore parasitic on, but not identical to, the predicate-offence investigation conducted by the police or CBI. Because the Act creates a self-contained code, the Supreme Court has repeatedly held that PMLA procedure prevails over the general CrPC wherever the two are inconsistent.

It is useful to fix at the outset the conceptual scaffolding that runs through every power discussed below. Each ED power is built on the same two-part design: an internal trigger, almost always a "reason to believe" that must be recorded in writing and grounded on "material in possession," and an external check, usually a duty to transmit that reasoning and material to an independent forum (the Adjudicating Authority or a court) within a defined time. The Vijay Madanlal Bench treated this recurring design as the constitutional justification for the breadth of the powers: because the agency must commit its reasons to writing and submit them to outside scrutiny, the power is, in the Court's view, structurally non-arbitrary. The post-2022 cases can be understood as the judiciary insisting that these paper safeguards be honoured in substance, not merely recited. For the statutory architecture and the FATF origins of these powers, see the genesis chapter and the broader PMLA notes hub.

The Power to Summon and Interrogate: Section 50

Section 50 is the workhorse of ED investigation. Sub-sections (2) and (3) empower the Director and authorised officers to summon any person whose attendance is considered necessary, to give evidence or produce records. The summoned person is bound to attend in person or through an authorised agent and is bound to state the truth on any subject of examination. Crucially, Section 50(4) deems every such proceeding to be a "judicial proceeding" within the meaning of Sections 193 and 228 of the Indian Penal Code, meaning a false statement can attract prosecution for perjury.

In Vijay Madanlal Choudhary, the Court characterised a Section 50 inquiry as being in the nature of an inquiry and not a criminal investigation in the conventional sense. Because the person summoned is not, at that stage, formally an accused, and because ED officers are not police, the Court held that statements recorded under Section 50 are not hit by the protection against self-incrimination in Article 20(3). This makes the summons power formidable: a witness must appear and answer truthfully, and his answers may later be used.

That breadth has limits. In Prem Prakash v. Union of India (2024 INSC 637), the Supreme Court held that where a person is already in custody (even in another matter) of the same investigating agency, a statement recorded from him under Section 50 cannot be considered the product of a free mind and is inadmissible against the maker. The Court reasoned that to hold otherwise would be contrary to fair play and to the protection in Article 20(3). Section 50 thus operates fully against a free witness but is blunted the moment custody removes the element of voluntariness.

The doctrinal pivot worth grasping is the moment of transition from witness to accused. A person who attends on summons as a witness enjoys no Article 20(3) shield, because that guarantee protects only a person formally "accused of an offence." But the protection crystallises once he is arrayed as an accused or taken into custody, and from that point the compulsion inherent in custody contaminates any statement extracted by the same agency. For examination purposes, the safe formulation is that Section 50 is not unconstitutional, but its product is admissible only so long as the maker was a free, un-accused witness when the statement was recorded. A summons issued purely to advance a continuing inquiry, even after a complaint has been filed against others, does not by itself infringe the right against self-incrimination; what matters is the status and the freedom of the particular maker.

Search and Seizure: Sections 17 and 18

Section 17 authorises the search of premises and seizure of property and records where an authorised officer has "reason to believe," recorded in writing, that a person has committed acts constituting money laundering or possesses proceeds of crime or relevant records. Section 18 confers the parallel power to search a person. The 2019 amendment deleted the earlier proviso that had tied a Section 17 search to the prior filing of a report under Section 157 of the CrPC for the predicate offence.

In Vijay Madanlal Choudhary, the petitioners argued that deleting this proviso unleashed arbitrary searches. The Court upheld the amended provisions, locating the safeguards in the structure of the section itself: the requirement of recorded reasons to believe, the obligation to forward the reasons and the material to the Adjudicating Authority, and the deterrent of penalties for vexatious search under the Act. The seized material must be placed before the Adjudicating Authority, which polices the legitimacy of the seizure. The practical limit, therefore, is documentary discipline: a search unsupported by contemporaneous written reasons grounded in tangible material is vulnerable to challenge.

Property and records seized under Section 17 do not remain indefinitely with the ED. Section 20 permits retention of seized property only where an authorised officer records, in writing, a reason to believe that it is required for adjudication, and even then the Adjudicating Authority must confirm retention. Retention without such confirmation beyond the statutory period is unsustainable.

Provisional Attachment Before Conviction: Section 5

Perhaps the most economically consequential power is the freezing of property under Section 5. The Director or an authorised officer may, on "reason to believe" recorded in writing, provisionally attach property suspected to be proceeds of crime. Such an order operates for a maximum of 180 days, within which the ED must file a complaint before the Adjudicating Authority, failing which the attachment lapses by operation of law.

Provisional attachment is civil and preventive, not punitive, and operates independently of any arrest or conviction. The Vijay Madanlal Bench upheld Section 5, again pointing to its inbuilt controls: recorded reasons, the 180-day outer limit, the mandatory reference to an independent Adjudicating Authority, and the appellate ladder to the Appellate Tribunal and the High Court. The limit on the power is thus temporal and procedural: attachment cannot be open-ended, and it cannot survive without timely confirmation. The first proviso to Section 5(1), which once required the prior registration of a predicate-offence report, and the second proviso, which allows attachment even without it where non-attachment would frustrate proceedings, together calibrate how early the ED may freeze assets.

Two features distinguish attachment from the criminal-process powers. First, it bites on property, not on the person, and so does not engage Article 22 or the bail regime at all; its constitutional touchstones are Articles 14 and 300A (deprivation of property by authority of law). Second, because it is a civil-preventive measure aimed at preserving the proceeds of crime pending adjudication, the standard of satisfaction is the agency's "reason to believe" rather than proof, and the property owner's remedy lies in the adjudication and appeal process rather than in resisting the attachment at its inception. The discipline that the case law imposes is, once again, temporal: the 180-day clock is mandatory, and an attachment not confirmed within the statutory scheme simply ceases to operate, returning the property to the person from whom it was taken.

The Power to Arrest: Section 19 and Its Internal Safeguards

Section 19(1) permits a designated ED officer who, on the basis of material in his possession, has "reason to believe" (recorded in writing) that a person is guilty of an offence under the Act, to arrest that person and, as soon as may be, to inform him of the grounds of arrest. Section 19(2) requires the arresting officer immediately to forward a copy of the arrest order and the underlying material, in a sealed envelope, to the Adjudicating Authority. Section 19(3) mandates production before the jurisdictional Magistrate or court within 24 hours.

In Vijay Madanlal Choudhary, the Court upheld Section 19, characterising these requirements (material in possession, recorded reasons, sealed-cover transmission, 24-hour production) as inbuilt safeguards ensuring non-arbitrariness and accountability. But the judgment left the contours of "grounds of arrest" and the scope of judicial review under-specified, and it is the post-2022 line of cases that has given Section 19 real teeth.

Three features of the section repay close attention. First, the threshold is "reason to believe," a higher bar than the "reasonable suspicion" that suffices for a police arrest under the CrPC; it demands a degree of proximity between the material and the conclusion of guilt. Second, the belief must rest on "material in his possession" at the time of arrest, so the ED cannot retrospectively justify an arrest by material gathered afterwards. Third, the sealed-cover transmission to the Adjudicating Authority is not a formality: it creates a contemporaneous, tamper-resistant record against which the legality of the arrest can later be tested. These three features are exactly the points at which the later cases bite, because each is a place where the agency's paperwork can be checked against the constitutional promise of Article 22.

Written Grounds of Arrest: Pankaj Bansal and Ram Kishor Arora

The decisive turn came in Pankaj Bansal v. Union of India (2023 INSC 866, decided 3 October 2023). A two-Judge Bench held that merely reading out or orally communicating the grounds of arrest does not satisfy Section 19 read with Article 22(1) of the Constitution. The grounds must be furnished to the arrestee in writing, as a matter of course and without exception, so that he can effectively challenge the arrest and seek bail. On the facts, the Court found the arrest non-compliant and set it aside.

The reach of Pankaj Bansal was promptly tested. In Ram Kishor Arora v. Directorate of Enforcement (2023 INSC 1082, decided 15 December 2023), the Court clarified two points: first, the written grounds must be furnished within 24 hours of arrest, not necessarily at the very instant of arrest; and second, the Pankaj Bansal requirement of written grounds operates prospectively from 3 October 2023 and does not retrospectively vitiate earlier arrests where grounds were otherwise conveyed. Together these decisions convert a soft "as soon as may be" obligation into a hard, time-bound, documentary one whose breach renders the arrest illegal.

The rationale links Section 19 to Article 22(1) of the Constitution, which guarantees every arrested person the right to be informed of the grounds of arrest and to consult and be defended by a legal practitioner of his choice. The Court reasoned that this right is illusory unless the grounds are reduced to writing: an oral recital is unverifiable, can be reconstructed after the fact, and leaves the arrestee unable to instruct counsel or frame a bail application with precision. The consequence is significant for the burden of proof. Where the ED cannot produce contemporaneous written grounds, the arrest is liable to be declared illegal and the arrestee released, regardless of the strength of the underlying allegations. Compliance is thus a condition precedent to a lawful arrest, not a curable irregularity, and it operates independently of whether the "reason to believe" was itself sound.

Beyond Belief: "Need and Necessity" After Arvind Kejriwal

If Pankaj Bansal policed the form of arrest, Arvind Kejriwal v. Directorate of Enforcement (2024 INSC 512, decided 12 July 2024) probed its substance. Justices Sanjiv Khanna and Dipankar Datta reaffirmed that an arrest under Section 19 is judicially reviewable: a court may strike down an arrest where the "reason to believe" is not founded on objective material or where the power has been exercised arbitrarily, though it should not conduct a merits trial at the arrest stage.

The Bench went further, framing as an open and substantial question whether "need and necessity to arrest" is a distinct, justiciable ingredient of a valid Section 19 arrest, over and above the existence of incriminating material, and referred that question for authoritative consideration. The signal is unmistakable: the ED cannot treat arrest as automatic once it forms a belief of guilt; the proportionality of depriving a person of liberty is itself open to scrutiny. This dovetails with the bail jurisprudence considered below and with the consequences of conviction explored in punishment for money laundering.

The Cognizance Cut-Off: Tarsem Lal

A further, structural limit on the arrest power emerged in Tarsem Lal v. Directorate of Enforcement (2024 INSC 434, decided 16 May 2024). The Court held that once a Special Court takes cognizance of a complaint filed under Section 44 of the PMLA, the ED's power to arrest under Section 19 is exhausted with respect to an accused who has not been arrested during the investigation. From that point the matter is governed by the post-cognizance procedure, and if the ED requires custody it must apply to the Special Court for remand under Section 309(2) of the CrPC rather than arresting under Section 19.

Equally significant, the Court held that an accused who appears in answer to a summons issued by the Special Court after cognizance is not in "custody," and therefore need not satisfy the twin bail conditions under Section 45 merely to secure bail or a bond at that stage. Tarsem Lal thus draws a clear temporal boundary: Section 19 is a tool of the investigative phase, not a lever to be pulled after the prosecution complaint has been judicially noticed.

Arrest, Remand and ED Custody: V. Senthil Balaji

Once an arrest is validly made, the question becomes how long and on what terms the ED may keep an arrestee in its custody. In V. Senthil Balaji v. State (decided 7 August 2023), the Supreme Court held that Section 167 of the CrPC, including the grant of custody to the investigating agency, applies to PMLA arrests and complements Section 19. ED (agency) custody is therefore permissible, subject to the magisterial controls of Section 167.

The Court read the 15-day ceiling on police/agency custody not as confined to the first 15 days alone but as a cumulative cap deployable across the whole 60 or 90 day investigation window, a reading that gives the ED meaningful interrogation time while preserving the outer statutory limit. The decision confirms that PMLA arrest is integrated into, rather than insulated from, the established remand architecture of ordinary criminal procedure.

For students, the takeaway is the interplay between Section 19 and Section 167 CrPC. Section 19 governs the legality of the arrest; Section 167 governs the legality of the detention that follows. An arrest may be valid yet the remand challengeable, or vice versa. Senthil Balaji also illustrates that non-cooperation with summons can, in an appropriate case, supply part of the material grounding a "reason to believe," though it can never be the sole basis, since the right to remain silent and the protections of Section 50 would otherwise be hollow. The judgment should be read alongside Tarsem Lal: Senthil Balaji tells us how custody works during investigation, while Tarsem Lal tells us that the Section 19 route to custody closes once cognizance is taken.

The ECIR Is Not an FIR

A recurring grievance is that the ED commences investigation by recording an Enforcement Case Information Report (ECIR) but refuses to supply a copy to the suspect. In Vijay Madanlal Choudhary, the Court held that the ECIR is an internal departmental document and is not equivalent to an FIR under the CrPC. Consequently, the ED is not obliged to furnish the ECIR to the accused; it is sufficient, at the stage of arrest, that the grounds of arrest are disclosed (a disclosure that Pankaj Bansal later required to be in writing).

This holding is one of the most criticised features of the PMLA edifice, since non-supply of the foundational document constrains the accused's ability to mount an early challenge. But it is settled law: the limit on the ED here is the disclosure of grounds, not the disclosure of the ECIR itself. The practical safeguard lies in the quality and contemporaneity of those recorded grounds, now subject to written communication and to judicial review of the underlying "reason to believe."

How Far Down the Chain? Pavana Dibbur and the Predicate Link

The ED's jurisdiction is anchored to "proceeds of crime" generated by a scheduled offence, so the reach of its powers depends on how loosely that link may be drawn. In Pavana Dibbur v. Directorate of Enforcement (decided 29 November 2023), the Court clarified two points of significance. First, a person need not be an accused in the predicate (scheduled) offence to be prosecuted for money laundering; what matters is involvement in a process or activity connected with the proceeds of crime. This widens the field of potential targets.

Second, and limiting the ED, the Court held that the offence of criminal conspiracy under Section 120-B of the IPC qualifies as a scheduled offence only where the object of the conspiracy is itself an offence listed in the Schedule. The ED cannot bootstrap jurisdiction by alleging a bare Section 120-B conspiracy to an unscheduled offence and thereby manufacture a predicate. The contours of "proceeds of crime" and the scheduled-offence link are developed further in the definitions chapter and the chapter on the offence of money laundering.

Bail as a Brake: Section 45 and the Twin Conditions

The most powerful practical lever the ED holds after arrest is the stringency of bail. Section 45 imposes "twin conditions": the court must be satisfied that there are reasonable grounds for believing the accused is not guilty and that he is not likely to commit any offence while on bail, before granting bail. In Nikesh Tarachand Shah v. Union of India (2018), the Supreme Court struck down the twin conditions as then framed for offending Articles 14 and 21.

Parliament responded through the Finance Act, 2018, recasting the trigger from "offences under Part A of the Schedule" to "any offence under this Act." In Vijay Madanlal Choudhary, the Court held that this amendment cured the defect identified in Nikesh Tarachand Shah and revived the twin conditions as constitutionally valid. The conditions therefore operate as a formidable limit on the accused's liberty, but the Court has tempered them in practice: in Prem Prakash it reiterated that "bail is the rule and jail is the exception" survives even under Section 45, and in Tarsem Lal it held that the twin conditions are not triggered where an accused merely appears on summons after cognizance.

The Emerging Balance: Power Constrained by Process

Read together, the statute and the case law describe a deliberate equilibrium. The ED's powers, summons under Section 50, search under Section 17, attachment under Section 5, arrest under Section 19, are extraordinary and front-loaded, exercisable on the agency's own "reason to believe" before any court has tested the allegation. Vijay Madanlal Choudhary sustained that architecture, treating each power's recorded-reasons-plus-oversight design as a sufficient safeguard.

But the years since have shown those safeguards being made justiciable rather than merely declaratory. Pankaj Bansal and Ram Kishor Arora hardened the grounds-of-arrest obligation into a written, 24-hour duty; Arvind Kejriwal opened "need and necessity" to review; Tarsem Lal switched off Section 19 at cognizance; Prem Prakash protected custodial statements; and Pavana Dibbur policed the predicate link. The ED's powers remain among the widest in Indian criminal enforcement, but they are now exercised under a thickening canopy of process, where a defect in documentation, timing, or jurisdiction can unravel the action. For aspirants, the examinable core is precisely this dialectic: identify the statutory power, name its internal trigger, and pair it with the case that supplies the external limit. The institutional machinery that adjudicates many of these limits is examined in the chapter on the Adjudicating Authority.