Open Access Research Article
White Black Legal – International Law Journal · ISSN 2581-8503
From Documentary Compliance to Transaction Verification: A Critical Analysis of Fraudulent Input Tax Credit under the GST Regime
Read the Full Research Paper
Access the complete open-access article in PDF format. No login is required.
Abstract
FROM
DOCUMENTARY COMPLIANCE TO TRANSACTION VERIFICATION: A CRITICAL ANALYSIS OF
FRAUDULENT INPUT TAX CREDIT UNDER THE GST REGIME
AUTHORED
BY - ADITYA NIGAM
National
Law Institute University, Bhopal
CO-AUTHOR
- ANSH SHUKLA
National
Law Institute University, Bhopal
Abstract
Input
tax credit (ITC) is the mechanism that makes the Goods and Services Tax a tax
on value added, and fraud on ITC is the system's most persistent vulnerability.
Indian law has long answered that vulnerability with a documentary
architecture: a credit is claimed on a tax invoice, matched through returns,
and supported by e-invoices and e-way bills. A recent press-reported
departmental notice, alleging that photographs and recycling certificates were
digitally manipulated to support credit claims for e-waste services,
illustrates what that architecture cannot do. A record can be valid on its face
yet correspond to no supply. This article asks whether the existing ITC
framework can detect and deter fraud supported by apparently valid but
fabricated electronic compliance material, and whether Indian law should move
towards risk-based transaction verification. Using doctrinal method, it
examines the Central Goods and Services Tax Act 2017, the Rules, CBIC
circulars, and decisions of the Supreme Court and High Courts, with particular
attention to State of Karnataka v Ecom Gill Coffee Trading Pvt Ltd. It
argues that Indian law runs together three distinct questions: whether a supply
occurred, whether tax reached the exchequer, and whether the recipient was
culpable. It also argues that the corroboration demanded in Ecom Gill is
itself documentary and so vulnerable to fabrication. The article proposes a
statutory framework of tiered, risk-triggered verification, a prescribed
evidentiary standard grounded in independence of provenance, a bounded safe
harbour, and procedural safeguards for algorithmic and electronic evidence.
Keywords:
Goods
and Services Tax; Input Tax Credit; Fake Invoices; Digital Tax Fraud;
Transaction Verification; Taxpayer Due Diligence; Electronic Evidence; GST
Compliance
I. Introduction
A
tax invoice is a statement about the world. It says that a supplier delivered
something to a recipient and charged tax on it. The Goods and Services Tax
(GST) treats that statement as the starting point for a credit that the
recipient can set against its own liability. The credit is not a favour at the
margin of the system. It is the mechanism by which GST taxes value added rather
than turnover, and it is the reason that a fraudulent invoice is worth money.[1]
The
reported scale of the problem is large. Parliamentary replies, as reported by
the press, record that central GST formations detected fake ITC of Rs 36,374
crore in 2023-24 across 9,190 cases, and that ITC fraud accounted for 15,283 of
the 30,056 evasion cases detected in 2024-25, with evasion of Rs 58,772 crore.[2]
Two cautions apply. These are detections and allegations, not adjudicated
findings, and the series differ in coverage: a figure of Rs 2.01 lakh crore
attributed to the Directorate General of GST Intelligence (DGGI) for 2023-24 in
one of the secondary papers consulted for this article could not be reconciled
with the Parliamentary series for central field formations, and I do not rely
on it.[3]
What can be said with confidence is that the enforcement apparatus itself
identifies fake ITC as a major and recurring category of fraud.
The
legal problem this article addresses is narrower than 'fake invoices'. Early
fake-invoice fraud was crude: shell entities, stolen identities, invoices for
nothing. The prevailing response was documentary. Registration was tightened,
invoices were matched against returns, e-invoicing and e-way bills were
introduced, and the recipient's entitlement came to depend, by statute, on
possession of the right documents. That architecture works against crude fraud.
It is far less effective against fraud that manufactures the documents the
architecture demands. The question is what GST law should do when apparent
compliance is itself the instrument of deception.
A
reported controversy supplies a contemporary illustration. On 30 September
2026, Reuters reported on an August 2026 departmental notice alleging that
three appliance manufacturers had claimed ITC on e-waste recycling services
that were never rendered, and that photographs and certificates offered as
proof of collection and processing had been manipulated.[4]
The notice has not, to my knowledge, been published, its allegations are
contested or awaiting reply, and no adjudication has occurred. I use the
episode as a case study of a structural problem, not as a source of conclusions
about any party.
The
article makes three arguments. First, Indian ITC law conflates three questions
that should be kept apart: whether a supply occurred, whether tax reached the
exchequer, and whether the recipient knew or should have known of a defect.
Different evidence, burdens and consequences belong to each. Second, State
of Karnataka v Ecom Gill Coffee Trading Pvt Ltd is the strongest judicial
authority for requiring proof of the transaction behind the invoice, but the
corroboration it demands (vehicle details, freight, acknowledgments of
delivery) is documentary and can be fabricated. The judgment therefore points
to a principle, namely independence of proof, that it does not itself develop.
Third, a shift towards transaction verification is justified only if it is
risk-based, rule-governed and accountable. Indiscriminate verification would
impose unbounded investigative duties on honest recipients and hand officers
opaque discretion. The article closes with a reform framework designed to keep
revenue protection, fraud prevention, commercial certainty and procedural fairness
in tension rather than in sequence.
The
method is doctrinal. Primary sources are statutes, rules, notifications,
circulars and judgments; secondary sources are used to identify questions and
comparative evidence, not as authority for propositions of law.
II. Conceptualising ITC and Fraud under
GST
A.
What the credit is, and why that matters for verification
Courts
have characterised credit in two ways that pull in different directions. In Eicher
Motors Ltd v Union of India, the Supreme Court treated Modvat credit as a
vested right once the input duty had been paid.[5] In
Jayam & Co v Assistant Commissioner, it treated ITC under the Tamil
Nadu VAT Act as a concession whose statutory conditions must be strictly
complied with.[6] The tension is real, but
for present purposes it can be put in a single proposition. The credit is a
legal entitlement that crystallises only when the statutory conditions are met,
and the conditions in section 16 are framed as facts about the world: the
recipient holds a tax invoice, has received the goods or services, and
the tax has been paid.[7] A
document can be proof of a fact; it cannot be the fact.
Section
155 makes the allocation of proof explicit: where a person claims to be
eligible for ITC, the burden of proving the claim lies on that person.[8]
The significance of this provision is often underestimated in discussion of Ecom
Gill, to which I return. The burden is a settled feature of the statute and
not an invention of the Supreme Court.
B.
A taxonomy of defective invoices
The
vocabulary of 'fake' invoices is loose in practice and the statute does not
define it.[9]
Because different consequences attach to different defects, the categories need
to be kept distinct.
A
genuine invoice records a supply that took place, between the persons
named, of the goods or services described, at the value stated. An invoice
without underlying supply records no supply at all, and is the core case of
ITC fraud. An inflated invoice records a real supply at an overstated
value or rate; the supply exists, and the defect lies in quantum or
classification. Circular trading involves invoices passing among related
or controlled entities without real movement of value or goods, generating credit
or turnover. Shell-company transactions involve invoices issued by
registrations that have no real business at the stated place. These last three
categories may overlap with the second, but they are analytically separate,
because in inflated-invoice cases a supply exists and a proportionate remedy is
possible, while in the pure no-supply case it is not.
Two
statutory categories cut across these. Wrongful availment is taking
credit without satisfying the conditions of entitlement; wrongful
utilisation is applying credit already in the ledger to discharge
liability.[10] Credit may be wrongly
availed yet never utilised; blocking, discussed below, targets precisely that
interval.
A
further category deserves recognition because the contemporary controversy
turns on it: digitally fabricated compliance documentation. By this I
mean electronic material (photographs, certificates, logs, GPS or weighbridge
records, e-way bills) created or altered to give the appearance of statutory or
regulatory compliance. It is distinct from a fake invoice in two respects. The
fabricated material may be corroborative rather than constitutive: the
invoice may be genuine as an instrument, while the evidence offered to show
that the underlying activity occurred is false. And the fabrication may take
place outside the GST system, in a parallel regulatory system such as extended
producer responsibility, and reach the tax system only because the recipient relies
on it.
C.
Why fraud is the system's characteristic risk
Economic
accounts of VAT fraud have long observed that credit mechanisms create
refund-like claims on the exchequer that can be fabricated at low cost, and
that fraud concentrates where the credit chain is long and the verification of
upstream payment is weak.[11]
Self-assessment compounds the risk: the tax authority depends on filings whose
accuracy it cannot verify at scale. The secondary literature on India frames
this through asymmetric information and moral hazard.[12] I
note this framing but place no doctrinal weight on it. For legal purposes, the
relevant point is simpler: any rule that makes entitlement turn on documents
invites investment in producing documents.
III. The Legal Framework Governing ITC
Verification
A.
Entitlement and its conditions
Section
16 conditions credit on possession of a tax invoice or other prescribed
document, receipt of the goods or services, payment of tax by the supplier
(directly or through credit), and the filing of the recipient's return, with a
time limit for availment.[13]
The receipt condition contains a deeming provision which is important for
services: receipt is deemed where goods are delivered to another person on the
recipient's direction, or where services are provided to any person on the
direction and on account of the recipient.[14]
That deeming rule becomes significant in the e-waste case study, where the
recipient's service was performed, if at all, on physical waste that the
recipient never handled.
Since
2022 the statute has reinforced the supplier-side link. Matching provisions
were recast and an invoice must appear in the supplier's statement of outward
supplies for the recipient to be entitled.[15]
This does a great deal against the pure shell-invoice scheme, because a
recipient cannot easily claim from a supplier who has not reported the supply.
It does nothing against a supplier who reports a fictitious supply.
B.
Records and the evidentiary base
Registered
persons must keep accounts and records at their principal place of business and
retain them for the period prescribed (seventy-two months from the due date of
the annual return for the relevant year, as the section presently stands).[16] A
tax invoice must be issued in the form and with the particulars prescribed, and
above specified turnover thresholds the invoice must be generated through the
e-invoice system, which assigns an Invoice Reference Number and a QR code.[17]
Movement of goods above a prescribed value requires an e-way bill.[18]
Electronic records have a statutory route into evidence in GST proceedings
under section 145, which makes certain computer printouts admissible subject to
conditions, and the relevancy of statements made on summons is addressed in
section 136.[19]
C.
Demand, penalty and prosecution
Three
distinct enforcement tracks follow a finding of wrongful credit, and the choice
among them is where the substantive question of culpability is decided.
Civil
recovery. For periods up to 2023-24, section 73 applies to
non-fraud cases and section 74 to cases of fraud, wilful misstatement or
suppression of facts to evade tax, with different limitation periods and
penalty ladders; for 2024-25 onwards, the Finance (No 2) Act 2024 inserted
section 74A, which applies a common notice limitation period and preserves the
fraud/non-fraud distinction as one of penalty.[20]
Interest runs under section 50.
Penalty
independent of demand. Section 122(1) lists contraventions
attracting penalty equal to the tax evaded or credit wrongly availed (with a
floor of Rs 10,000). Two clauses are directly relevant: clause (ii), issuing an
invoice without supply, and clause (vii), taking or utilising credit without
actual receipt of goods or services. Section 122(1A) extends penalty to a
person who retains the benefit of such a transaction and at whose instance it
was conducted.[21] The CBIC has clarified
how these provisions operate on a chain of fake invoices: the supplier who
issues an invoice without supply cannot be subjected to demand of tax because
no supply occurred, but is liable to penalty under section 122(1)(ii); a recipient
who avails credit without receipt contravenes section 16(2)(b) and is exposed
to demand under section 74 together with interest and penalty.[22]
Prosecution.
Section 132(1)(b) and (c) criminalise, respectively, issuing an invoice without
supply leading to wrongful availment or utilisation of credit, and availing
credit on such an invoice. Imprisonment of up to five years attaches where the
amount exceeds Rs 500 lakh, and offences above that threshold are cognizable
and non-bailable.[23]
Section 135 creates a presumption of culpable mental state in prosecutions,
rebuttable by the accused, and requires facts to be proved beyond reasonable
doubt before the court treats the mental state as negated.[24]
Arrest under section 69 requires the Commissioner to have 'reasons to believe'
that the offence has been committed, and in Radhika Agarwal v Union of India
the Supreme Court upheld sections 69 and 70 while holding that the safeguards
of criminal procedure apply and that reasons to believe must rest on material,
be recorded, and be capable of challenge by the arrestee.[25]
D.
Interim and administrative measures
Three
measures operate before adjudication and deserve attention because they impose
the practical burden of fraud suspicion on the taxpayer. Under section 67,
officers may inspect, search and seize on 'reasons to believe'. Rule 86A
permits the proper officer to block the electronic credit ledger where he has
reasons to believe that credit has been fraudulently availed, including on
invoices issued by non-existent persons or without receipt of goods or
services; and section 83 permits provisional attachment.[26]
Registration may be suspended or cancelled, including where it was obtained by
fraud, wilful misstatement or suppression of facts.[27]
Separately, special all-India drives against suspected fake registrations have
been mounted using analytics derived from GSTN data. CBIC Instruction
02/2024-GST, for example, directed a drive from 16 August to 15 October 2024,
in which GSTN, with the Directorate General of Analytics and Risk Management,
would identify high-risk GSTINs from risk parameters, to be supplemented by
tools such as BIFA, ADVAIT, NIC Prime and e-way bill analytics and by human
intelligence.[28]
E.
The structural observation
Read
as a whole, the framework has a recognisable shape. It protects the exchequer
at three moments: at registration, through supplier-side verification; at
return, through matching; and after the event, through demand, penalty and
prosecution. What it largely lacks is a prescribed standard for proving the
transaction at the moment of claim or on a verification request. That gap
is filled by judicial elaboration of the section 155 burden, which is where Ecom
Gill enters.
IV. From Fake Invoices to Digitally
Fabricated Compliance
A.
What digitalisation changed
GST
digitalised both compliance and the evidence of compliance. Invoices pass
through an e-invoice portal; return data flow into an auto-generated statement
of available credit; e-way bills record vehicle, route and consignment. For the
tax administration these are valuable, because each document is time-stamped,
structured and machine-readable, and because inconsistencies between them can
be detected at scale. The international evidence on e-invoicing is cautiously
favourable. A study of Italy's mandatory e-invoicing found a significant
reduction in cross-border VAT fraud, measured through discrepancies in
double-reported trade data, and a study of Peru found improvements in
compliance.[29]
Digitalisation
also changed the economics of forgery. A paper invoice was hard to forge
convincingly at scale. An e-invoice is accepted by the portal if its structure
validates and its identifiers are consistent; the portal confirms that the
document is well-formed and properly authenticated, not that the transaction
behind it occurred. Press commentary has made the same point in the Indian
context: where both the e-invoice and the e-way bill are manufactured, the pair
can pass as a genuine transaction on the portals.[30]
The CBIC's own instruction on fake registrations tells officers to supplement
analytics with 'human intelligence', which implicitly concedes the limits of
system data.[31]
B.
The shift from fabricating instruments to fabricating evidence
Three
generations of fraud can be distinguished. The first fabricated the instrument:
an invoice from a non-existent supplier. The second fabricated the transaction
trail: shell entities with bank flows, e-way bills with invented vehicles,
circular payments. The third, which the e-waste episode exemplifies, fabricates
the evidence of performance: photographs of laden and empty trucks,
certificates of processing, correspondence. Each generation answers the
verification demands made of the previous one. When the law requires an e-way
bill, e-way bills are produced; when the courts require proof of movement,
proof of movement is produced. This is the central dynamic of the article. A
verification regime that specifies in advance which documents suffice will be
met by documents of that description.
V. The Contemporary E-Waste Documentation
Controversy
A.
What has been reported
The
facts below are drawn from press reports of a notice that has not been made
public, and they are allegations by tax authorities. Reuters reviewed a notice
dated 27 August 2026, running to some 550 pages, sent to Haier India, Godrej
& Boyce and Blue Star.[32]
According to that report, the notice alleged that the three companies had
wrongly claimed tax credits totalling about US$714,000 (Haier US$367,900,
Godrej US$159,700 and Blue Star US$186,600) on recycling services not in fact
rendered, and that the compliance certificates had been issued by a recycler,
WEEE-PRO Resource Recovery Solutions, which the notice says performed no
e-waste work. The notice is reported to say that the underlying procurement,
receipt and processing of e-waste is not established, that photographs said to
show trucks laden and then emptied were digitally altered, and that one
photograph time-stamped 29 minutes after sunset showed daylight and natural
shadows. It is further reported to say that the three companies admitted that
they had not independently verified the actual collection, transportation,
receipt, dismantling or recycling.[33]
The
responses reported are different in kind. Godrej & Boyce and Blue Star said
they would respond to the notice and were committed to legal compliance; they
pointed out that the recycler had been authorised by the government. Blue Star
said it had no reason to believe any irregularity existed but had stopped
dealing with the vendor. Haier India said it would continue to cooperate.
WEEE-PRO did not respond to Reuters, but the notice itself reportedly records
that its officials denied wrongdoing and disputed the evidence; an executive is
reported to have said that he edited photographs by 'adding a few clouds' but
that this was not for commercial use.[34]
Reuters adds that tax authorities are widening the inquiry across the industry,
and that the episode follows a January government warning about fraudulent
certification by spurious recyclers.[35]
Three
observations about the evidence of this record are in order. The document is a
notice which, on the reports, appears to be a show-cause notice and not an
adjudication order, and the companies had said only that they would respond.
The press report does not identify the provisions invoked, so any statement
about the legal basis of the notice is inference. And the statements attributed
to the recipients' admissions and to the recycler's executive are as reported
by Reuters from the notice; the recycler's executive disputes the evidence, as
the notice itself records.
B.
Legal issues that may become relevant
Without
suggesting what the notice invokes, the factual allegations engage a
recognisable cluster of provisions. If the recycling services were not
rendered, section 16(2)(b) (receipt of services) is in issue, as is the deeming
rule for services provided to others on the recipient's direction. If the
recycler is found to have issued invoices without supply, sections 122(1)(ii)
and 132(1)(b) are potentially relevant to the supplier, and sections
122(1)(vii) and 132(1)(c) to the recipient, though the latter depends on proof
of the requisite mental state. Whether proceedings would proceed under sections
73/74 or section 74A depends on the financial year of each claim. Reuters notes
that penalties for false claims can run up to 100 per cent; that ceiling corresponds
to the fraud limb of the demand provisions or to penalty under section 122, but
it is a general statement and not a description of what the notice proposes.
There
is also a logically prior question that the reports do not resolve: what
exactly was supplied? If the invoiced supply was a recycling service,
the relevant performance is physical processing of waste. If it was the issue
of a regulatory certificate, the character of the supply, and the proof that it
occurred, may differ. The point is not academic. The statutory condition for
credit is receipt of goods or services, and a regulatory certificate
that is not itself a supply of anything would be a poor foundation for credit
even if genuine.
C.
Evidence: how fabricated material could be established
The
notice, as reported, relied on several types of proof: examination of messaging
records, comparison of photograph time-stamps against sunset data, analysis of
images for alteration, and the statements of persons concerned. These are the
kinds of evidence that belong in a proceeding concerned with fabricated
electronic material. Part VIII considers what standard should govern them. Two
points arise at once. A time-stamp or metadata inconsistency is evidence that a
particular item is unreliable, but it does not establish that no activity
occurred; the Revenue still has to show that the underlying activity was not
performed, ordinarily through independent evidence such as the absence of
facility records, throughput capacity, power consumption, vehicle movement data
and bank flows. And an admission that the recipient did not independently
verify performance is an admission of non-verification, not of knowledge
of fabrication. The legal significance of non-verification depends on the
standard of diligence the law requires, which is exactly what this article
contends is unspecified.
D.
What the controversy reveals
The
reported facts illustrate five structural weaknesses.
First,
regulatory attestation substituted for tax verification. The recipients,
on their reported account, relied on a government authorisation held by the
recycler. That is a rational reliance for a regulatory purpose, but
registration under one regime says little about whether a particular supply
occurred for the purposes of another.
Second,
invisibility of performance. A purchaser of a service performed on
someone else's physical assets cannot observe performance. The Ecom Gill-type
evidence (vehicle, freight, delivery acknowledgement) is designed for goods
that the buyer receives; it fits awkwardly where the service is performed
upstream of the buyer.
Third,
corroboration sourced from the supplier. The photographs and
certificates came from the party whose performance was in question.
Corroboration generated by the person whose conduct is under examination is
evidence of what that person wishes to show.
Fourth,
the absence of a prescribed diligence standard. The reported complaint
is that the companies did not verify; the statute does not say what they were
obliged to verify. A standard defined after the event is a standard applied
retrospectively.
Fifth,
the gap between regulatory and fiscal systems. Tax authorities,
apparently, must reconstruct from first principles information that a regulator
for extended producer responsibility may hold or could hold: certificate
generation records, facility registration, throughput. An integrated
verification architecture would use it.
VI. Judicial Approaches to Transaction
Genuineness and ITC
A.
The decision in Ecom Gill
State
of Karnataka v Ecom Gill Coffee Trading Pvt Ltd
decided a group of appeals concerning input tax credit under section 70 of the
Karnataka Value Added Tax Act 2003.[36]
Section 70(1) put the burden on a dealer to prove that a claim to deduction of
input tax is correct, and section 70(2) imposed a penalty on a dealer who
knowingly produced a false invoice or other document to support such a claim.
In the lead matter, the assessing officer had denied credit because the selling
dealers had either cancelled their registrations or filed nil returns; in the
Ecom Gill matter, of 27 sellers, some were de-registered, some had not filed
returns, and some had denied turnover. The Tribunal and the High Court had
allowed credit, on the footing that the buyers held invoices and, in some
cases, had paid by account payee cheque.
The
Supreme Court reversed. Three propositions can be extracted. First, the burden
of proving the correctness of the claim lies on the purchasing dealer and
cannot shift to the revenue; a claim to be a bona fide purchaser does not
suffice.[37] Second, mere production
of invoices or evidence of payment by cheque is not enough; the dealer 'has to
prove beyond doubt the actual transaction', which may be shown by the name and
address of the selling dealer, details of the vehicle that delivered the goods,
payment of freight, acknowledgement of delivery, tax invoices and payment
particulars, and actual physical movement of the goods is sine qua non.[38]
Third, the statutory requirement to issue and retain a tax invoice under the
Rules is only one of the documents needed to discharge the burden of showing
genuineness.[39] The Court distinguished
the Delhi High Court's decision in On Quest Merchandising, which had
read down a provision denying credit for the seller's non-deposit of tax, on
the ground that the burden of proof was not in issue there.[40]
B.
What the judgment does not hold
Careful
reading limits the case in three ways that bear on its transposition to GST.
The
factual predicate. The assessing officer had recorded
reasons for doubting genuineness on the basis of evidence, notably cancelled
registrations and sellers' denials of the sales. The Court emphasised that in
none of the cases did the purchasing dealers produce any further supporting
material.[41] The case does not
authorise doubt without a basis; it concerns what is required of a dealer once
reasoned doubt has been raised. This matters for any reform, because it
suggests a threshold: it is the Revenue's showing of articulable red
flags that shifts the practical evidentiary burden.
The
standard of proof. The phrase 'beyond doubt' is imprecise.
Civil and tax adjudication normally proceeds on a preponderance of
probabilities, taking account of surrounding circumstances and human
probabilities.[42] Read literally, 'beyond
doubt' would impose a criminal standard on the taxpayer, which is hard to
reconcile with the statutory scheme. The better reading is that the Court was
describing the quality of material required, not setting a higher
standard of proof than the law ordinarily demands.
The
supplier-default issue. The judgment does not decide that a
recipient loses credit merely because the seller did not pay tax. Commentators
have noted that the Court distinguished rather than overruled On Quest,
and the Delhi High Court's reading down of the provision in that line of
authority, which turned on the distinction between bona fide purchasers and
others, was left standing by the Supreme Court's dismissal of the revenue's
special leave petition in Arise India.[43]
C.
Context and relevance: KVAT and CGST
Ecom
Gill
was decided on section 70 of the KVAT Act. Its holding about discharge of the
burden rests on the text of that provision, and the Court did not purport to
interpret the CGST Act. The statutory framework in which it was decided was a
State VAT with no return-matching architecture of the GST kind, and the
assessment years in the matters before the Court included 2010-11 and 2012-13.
The transposition to GST therefore requires an argument, not an assumption.
The
argument is available, and it has two parts. First, CGST section 155 is
functionally equivalent to section 70(1) on the allocation of burden, as
commentators noted immediately after the decision.[44]
Second, the CGST Act's condition that the recipient must have received
the goods or services (section 16(2)(b)) is a statutory basis for demanding
proof of actual receipt, and the Circular on fake invoices treats absence of
receipt as a contravention of that condition.
High
Courts have begun to apply the decision in GST cases. The Allahabad High Court,
in Anil Rice Mill v State of Uttar Pradesh, upheld proceedings under
section 74 where the taxpayer produced invoices, e-way bills and bank payments
but not freight payment, delivery acknowledgments or toll receipts, relying on Ecom
Gill and its own earlier decision in Shiv Trading.[45]
Equally, the same court has distinguished Ecom Gill where actual
movement of goods and return compliance were established.[46]
What is emerging is a fact-sensitive application, not a mechanical rule.
D.
The deeper doctrinal significance: documents about documents
The
enduring contribution of Ecom Gill is conceptual. It separates two
things that earlier case law, and some Tribunals, had merged: the documentary
validity of an invoice and the genuineness of the transaction it
records. A valid invoice is some evidence of a transaction, perhaps strong
evidence in ordinary circumstances, but not conclusive evidence. This is a
significant proposition and, in my view, correct. The statutory conditions for
credit are conditions about facts.
But
the list of evidence that the Court offers shows the limits of the approach.
Vehicle details, freight receipts and acknowledgments of delivery are
themselves documents. They are corroborative, and in a world where invoices are
forged they will be forged too. The judgment thus resolves one layer of the
problem (invoice alone is not enough) by retreating one step along an infinite
regress (invoice plus further documents). In an age of cheap digital
fabrication, further documents from the same party add little. What gives
corroboration its value is not the volume of paper, but its independence:
whether it originates from a source other than the person whose conduct is in
question, whether it was generated contemporaneously by a system outside that
person's control, and whether it can be cross-checked against other independent
records. This suggests that Ecom Gill provides a foundation for
transaction-level verification only if its reasoning is developed through the
criterion of independence of provenance. I develop that criterion in Part XI.
VII. Documentary Compliance versus
Transaction Authenticity
A.
Three questions the law runs together
Credit
disputes ordinarily involve three distinct factual and legal questions.
Did
a supply take place? This concerns the existence of a
transaction, for which the relevant statutory text is section 16(2)(b) and, at
a deeper level, the definition of supply. If no supply occurred, no input tax
was borne by the recipient, and a credit would be a gift from the exchequer.
Bona fides are irrelevant to this question; an innocent recipient of an invoice
for nothing has still received nothing.
Did
the tax reach the exchequer? This concerns the
supplier's compliance and engages section 16(2)(c) and the matching provisions.
A supply may be real while the supplier fails to pay. Here the recipient's bona
fides are relevant, as the Delhi High Court recognised in the VAT
context and as the EU courts have recognised for a longer time.
Was
the recipient culpable? This concerns knowledge,
recklessness or negligence, and bears on penalty, prosecution and the
characterisation of the case as one of fraud (sections 74, 122(1)(vii), 132).
The evidentiary question is different: not what happened but what the recipient
knew or should have known.
Indian
law blends these. A recipient who cannot prove the first (for example because
the documents offered are suspect) may be treated as culpable on the third,
because the same facts supply the red flags. A recipient who proves the first
but not the second may lose credit for the supplier's default. The consequences
differ markedly: loss of credit is a restitutionary consequence, penalty and
prosecution are punitive. Treating the same factual shortcoming as a basis for
all three is what generates both the appearance of arbitrariness for honest
taxpayers and the temptation to over-allege fraud.
B.
The case for transaction verification, and its costs
The
case for moving towards verification of the transaction rests on the
fabrication dynamic described above. If documents can be produced without
transactions, entitlement cannot rationally be tied to documents alone.
Verification at the level of the transaction targets the fact that the statute
makes decisive.
The
costs are equally real. Verification is expensive for taxpayers: small and
medium businesses cannot audit the premises of their suppliers. It is expensive
for the administration: field verification does not scale to the volume of
invoices. It creates discretion: an officer who may demand 'further
proof' of a transaction has a flexible instrument which can be used to extract
compliance or payment. And it unsettles commercial certainty: businesses rely
on the assumption that a validly invoiced purchase from a registered supplier
yields credit. The earlier Delhi High Court reasoning in Arise India
captures the point; an 'onerous burden' on a purchaser who has done all that
the law requires, and who has no control over the seller, treats the honest and
dishonest alike.[47]
The
relevant question is therefore not whether to verify but how much, by whom,
on what trigger, and with what consequences. A regime that verifies
everything burdens everyone; a regime that verifies nothing beyond the
documents invites fabricated documents. Neither is defensible. The conclusion
of this analysis is that verification must be proportionate to risk, and
that the allocation of verification between taxpayer and administration should
reflect who can do it most cheaply and reliably. The administration holds
information that the recipient does not (cross-supplier data, bank flows
reported to other authorities, registration histories); the recipient holds
information the administration does not (the commercial context, the physical
receipt, correspondence). A rational scheme assigns each what it can reasonably
know.
VIII. Digital Evidence and
Technology-Based Verification
A.
The evidentiary status of electronic compliance material
The
Bharatiya Sakshya Adhiniyam 2023 (BSA), in force since 1 July 2024, replaced
the Indian Evidence Act 1872.[48]
Three features matter here. Electronic and digital records may not be denied
admissibility merely because they are electronic (section 61), electronic
records in prescribed circumstances are treated as primary evidence (section 57
and its Explanations), and where a statement is to be given in evidence from
computer output, section 63 requires a certificate in the form set out in the
Schedule, which asks for the hash value of the record and is signed by the
person in charge of the device and by an expert.[49]
Two
conclusions follow for GST proceedings. First, the admissibility of a
GSTN record, an e-invoice or a recycler's photograph is rarely the real issue.
The real issue is weight: whether the record is reliable proof that the
thing it depicts occurred. Admissibility rules say a file is a document; they
do not say it is true. Second, the certificate regime points in a useful
direction because it requires hash values, which allow the integrity of a file
as produced to be demonstrated. It addresses tampering after the file is
secured, not fabrication before it, so a hash proves that the altered
photograph is the same altered photograph, not that the scene was real. It is
generally accepted that quasi-judicial adjudication is not bound by the
technical rules of evidence that govern trials, and I assume the same of the
BSA; sections 136 and 145 of the CGST Act provide separate gateways for
computer printouts and for statements made on summons, but the policy of the
BSA, that electronic evidence should be accompanied by technical assurance of
integrity, is a sensible reference point for the standard to be expected of the
Revenue when it alleges fabrication.[50]
B.
What standard should the Revenue meet when alleging fabrication?
When
a tax authority alleges that apparently authentic electronic compliance
documentation does not correspond to a genuine transaction, three layers of
proof are involved, and the standard of evidence should rise with the gravity
of the consequence.
Layer
one: that the item is unreliable. Forensic findings that a
photograph has been altered, or that its metadata is inconsistent with the
scene (the reported sunset comparison is an example), establish unreliability.
The method should be disclosed, the original files produced, and an expert
opinion given in a form that allows testing. The provisions on expert opinion
under the BSA, and the notified Examiner of Electronic Evidence under section
79A of the Information Technology Act 2000, supply an existing institutional
route.[51]
Layer
two: that the underlying activity did not occur.
This does not follow from layer one. A poorly doctored photograph of a real
delivery is still a poor photograph of a real delivery. The Revenue should show
absence of the activity through independent evidence: facility capacity,
utilities, vehicle movement records, financial flows, the statements of third
parties. The standard is the ordinary civil standard of the preponderance of
probabilities, applied with attention to surrounding circumstances.[52]
Layer
three: that the recipient knew or should have known.
For penalty under the fraud limb, for section 122(1A) and for prosecution, the
state of mind of the recipient must be established; the presumption in section
135 operates only in prosecutions, and even there it is rebuttable and does not
dispense with the prosecution's burden to establish the foundational facts. For
penalty, the Supreme Court has long held that a penalty ordinarily should not
be imposed for a technical or venial breach in the absence of deliberate
defiance or dishonest conduct.[53]
Sections 122(1)(vii) and 74 in their terms invite the question whether
recipient culpability is an element, and the law would be clearer if the
statute said so.
Third-party
statements raise a further difficulty. Where the case depends on a statement of
the supplier or its officers (as the reporting suggests in the e-waste matter),
the Supreme Court's insistence that denial of an opportunity to cross-examine
witnesses on whose testimony an adjudication depends can vitiate the order is
directly engaged.[54]
C.
Technology-based verification: promise and constraints
The
administration already uses analytics. CBIC and GSTN deploy tools such as BIFA,
ADVAIT and NIC Prime to generate risk profiles, and instructions direct
officers to combine them with human intelligence.[55]
Both of the secondary papers consulted treat advanced analytics,
risk-profiling, invoice matching and artificial intelligence as part of the
answer.[56] I
agree with the direction but want to set out the legal conditions of
acceptability.
The
analytic techniques have distinct uses. Invoice and three-way matching
(invoice, e-way bill, payment) tests internal consistency. Network analysis
identifies shared directors, addresses, IP addresses and phone numbers: the
Delhi State GST department's investigation of a Rs 347 crore fake-billing
scheme, for example, relied on shared IP addresses, contact numbers and email
addresses among firms.[57] Risk
scoring ranks taxpayers for scrutiny. Integration of financial and tax
data tests whether the declared business is consistent with bank and
customs flows.
Each
is subject to constraints that law must address. A risk score is a prediction,
not a finding; false positives will fall on honest taxpayers, and where the
score triggers blocking of credit under rule 86A or issuance of a notice, the
cost of a false positive is borne at once. Algorithmic opacity makes it hard
for a taxpayer to know what is alleged. Data quality is uneven, and a mismatch
may reflect a clerical error. Privacy interests are engaged by the integration
of bank, tax and other personal data, and the constitutional standard of
proportionality discussed below applies.[58]
The upshot is that technology belongs in the selection and investigation
of cases, not in their decision. A flag from a model is a reason to
look, never a reason to find.
D.
What the comparative experience adds
The
EU experience is instructive on two points. On the recipient's position, the
Court of Justice has held since Kittel that a taxable person who knew or
should have known, by his purchase, that he was participating in a transaction
connected with fraudulent evasion of VAT loses the right to deduct; in Mahagében
and Bonik it held that the authorities may not deny deduction because of
upstream fraud unless they establish, on the basis of objective evidence, that
the taxable person knew or should have known.[59]
The Court's approach in Mahagében has been summarised as holding that a
tax authority cannot, as a general rule, require a trader to verify that its
supplier has filed returns and paid its VAT. These authorities have no
counterpart in the Indian statute, but they offer a model of how to separate
the question whether a supply is genuine from the question whether the
recipient is to blame. On digital administration, the EU's VAT in the Digital
Age package, adopted in 2025, introduces digital reporting requirements based
on structured e-invoicing for intra-EU transactions, and the evidence from
Italy suggests that mandatory e-invoicing reduces cross-border fraud, though
the studies concern a different kind of fraud (missing-trader fraud in
cross-border trade) and the measure of success is indirect.[60]
The comparative lesson for India is not that e-invoicing is sufficient, but
that its benefits depend on administrative follow-up through audit.
IX. Taxpayer Due Diligence and Recipient
Liability
A.
The scope of the question
How
far should a recipient be required to verify the authenticity of a transaction
before claiming credit? The question has an easy answer at the extremes. A
recipient who knowingly participates in a fraud should not receive credit. A
recipient who is required to investigate every supplier of every input cannot
function. The difficulty lies in the middle, and it is made harder by an
asymmetry: the Revenue defines diligence after the event, with the benefit of
knowing the fraud, while the taxpayer must decide before.
B.
A spectrum of diligence
Diligence
measures can be arranged by cost and by what they establish. Supplier
verification (checking GST registration status and filing history on the
portal) is cheap, standard and proves little about a given transaction. Invoice
and e-way bill verification confirms the document trail. Payment
verification (through banking channels) is evidence that money moved,
though circular flows can fabricate this too. Physical receipt records
(goods inwards register, weighbridge slips, inventory movement) are within the
recipient's own control, are reasonably cheap and strongly corroborate receipt
of goods. Transport documentation adds an independent source when it
comes from a third-party carrier or toll system. Supplier capacity (does
the supplier have the premises, equipment, staff and power to produce what it
invoiced?) is relevant but costly; it is ordinarily proportionate only for
large or unusual purchases. Commercial correspondence and third-party
documentation have value in proportion to their independence.
On
this spectrum the distinction the paper's argument requires can be drawn. Reasonable
diligence is the set of measures a prudent business in the recipient's
position would take, proportionate to the value, novelty and risk
characteristics of the transaction, and verifiable from records in the
recipient's own possession. An unrealistic obligation would require the
recipient to audit the supplier's own operations, or those of the supplier's
suppliers, or to guarantee the supplier's tax payment. The first is an
acceptable condition of an entitlement which the state funds. The second
converts the taxpayer into an unpaid tax inspector.
C.
Receipt of services performed on third-party assets
The
diligence question takes on a distinct form for services whose performance the
buyer cannot observe: recycling, certification, compliance consulting,
marketing, back-office services. For goods, physical receipt is observable. For
services, the buyer's evidence of receipt is the output (a report, a
certificate, a deliverable) and the buyer's ability to verify performance
depends on the service's nature. A rational diligence standard for such
services should be risk-based, requiring more of the buyer where performance is
invisible, where the supplier is new or thinly capitalised, where payment is
out of proportion to the work, and where the buyer's own regulatory interest in
the output (as in compliance certificates) creates an incentive to accept it
without scrutiny. That last factor needs care, because it is the point at which
a purchaser's diligence becomes a question of reasonable suspicion, and it is
why regulatory-compliance purchases should be placed in a higher-risk tier than
ordinary inputs.
D.
Protection for the bona fide recipient
Should
the law give greater protection to recipients who have complied with prescribed
requirements? The Arise India line says that denying credit to bona fide
purchasers for the seller's default fails to distinguish between the diligent
and the complicit and so offends equality. The reasoning supports a statutory
safe harbour, but the safe harbour must be shaped carefully.
A
safe harbour cannot preserve credit where no supply occurred. The
condition of receipt is a condition about facts, and an innocent recipient of a
fabricated invoice has no entitlement to the credit, though it may well have a
claim against the supplier. What a safe harbour can do, legitimately, is
(i) preserve credit where a supply is established but the supplier has
defaulted on payment of tax, (ii) exclude penalty and prosecution where the
recipient complied with the prescribed diligence and had no knowledge, and
(iii) fix the date from which a new diligence standard applies. This division
respects the structure of the three-questions analysis in Part VII.
X. Constitutional and Procedural
Constraints
A.
Article 14: arbitrariness and classification
The
strongest constitutional objection to a verification regime is that it may be
arbitrary. A provision or practice that treats honest and dishonest purchasers
alike lacks a rational basis, as the Delhi High Court held in the VAT context.
An unstructured discretion to demand 'further proof' is also vulnerable,
because it permits unequal treatment of similarly situated taxpayers without
any criterion. The remedy is not to forbid verification but to require that it
be rule-governed: criteria published, triggers stated, treatment consistent.
The same principle supports risk-based verification, since classification by
risk factors is a rational basis if the factors have a demonstrable
relationship to fraud.
B.
Article 19(1)(g) and proportionality
Restrictions
on the right to carry on business must be reasonable, and the Supreme Court's
proportionality analysis asks whether the measure has a legitimate aim, is
rationally connected to it, is necessary in that no less restrictive
alternative would achieve it equally, and strikes a fair balance.[61]
Applied to verification, the necessity limb is critical: if a risk-based,
targeted measure would achieve equivalent protection at lower cost to honest
taxpayers, an undifferentiated duty is disproportionate. That is the
constitutional foundation for preferring risk-based to universal verification.
C.
Article 21 and personal liberty
Verification
may be preliminary to prosecution and arrest. Radhika Agarwal
establishes that the powers under sections 69 and 70 are valid, that arrest
must rest on material and recorded reasons to believe, and that the arrestee is
entitled to the grounds. A finding that a recipient 'failed to verify' cannot
by itself supply reasons to believe that the recipient committed an offence
requiring a culpable mental state, and prosecution should not be a default
consequence of an inadequate verification file. This article's call for a
graduated response to non-verification is consistent with that holding.
D.
Article 265, delegated standards and legitimate expectation
Article
265 provides that no tax shall be levied or collected except by authority of
law. Denial of credit on the ground of failure to meet a diligence standard not
found in the Act or Rules, and instead articulated in circulars, instructions,
or the elaboration of a judgment, raises a question of authority. Circulars
issued under section 168 bind the department; they cannot enlarge the statute
against the taxpayer.[62]
If a diligence standard is to be a condition of entitlement, it should be in
the Rules.
Legitimate
expectation points the same way. The assessment years in Ecom Gill were
2010-11 and following; the standard was declared in 2023. A taxpayer who bought
in reliance on invoices and registration checks as then understood has a
plausible expectation which the law should respect, particularly with penalty
and prosecution.[63] A new evidentiary
standard should operate prospectively, at least as regards penal consequences.
E.
Natural justice, reasons and review
Procedural
fairness has specific demands. Section 75 requires a hearing where adverse
decision is contemplated and requires the order to set out the relevant facts
and the basis of decision.[64]
The Supreme Court has treated reasons as an essential component of fair
decision-making and has said that a breach of natural justice should be
established by showing prejudice.[65]
For verification, this means an order denying credit on the footing that proof
is insufficient should say what was required, why what was produced was
inadequate, and what evidence the Revenue relied on, including any algorithmic
indicator. Appeal lies under the Act, and the tribunal stage has, historically,
been delayed, which compounds the cost of an erroneous initial decision.[66]
XI. Towards a Risk-Based Transaction
Verification Framework
A.
Design premises
The
framework proposed here rests on five premises drawn from the preceding
analysis. Entitlement turns on a fact (supply), so some transaction-level proof
is legitimate. Corroboration is valuable in proportion to its independence, so
the law should specify the type of proof, not the quantity of paper. The cost
of verification should be allocated to the party best placed to bear it.
Culpability must be established separately from entitlement, with penal
consequences confined to knowing or reckless conduct. And every power to verify
must be accompanied by reasons, review and a record.
B.
Proposals
Each
proposal is stated in the form: what should change, why, how it should operate,
and what safeguards are needed.
1.
A prescribed evidentiary standard for
proof of supply. What: amend the CGST Rules to
prescribe, for goods and for services, categories of evidence that suffice to
discharge the section 155 burden, graded by independence of provenance. Why:
Ecom Gill leaves the content of the burden to case-by-case elaboration,
and the specified items are documentary and forgeable. How: a rule under
section 16 (or a new sub-section authorising rules) that classifies evidence
into (a) system-generated or third-party records outside the claimant's control
(bank records, toll or tracking data, carrier records, regulator portal
records), (b) claimant-generated contemporaneous business records (inwards
registers, stock ledgers), and (c) supplier-generated documents. Category (c)
alone does not suffice; credit should be supported by (a) or by (b)
corroborated by (a) at the level appropriate to the risk tier. Safeguards:
the Rule should provide that no document of a type not prescribed may be
required, and that the claimant may offer alternative equivalent proof.
2.
Three tiers of verification,
triggered by stated indicators. What: replace an
undifferentiated demand for proof with tiers. Why: proportionality under
article 19(1)(g), and the cost asymmetry between small and large buyers. How:
in the baseline tier, ITC supported by a valid e-invoice, matched GSTR-2B entry
and e-way bill (for goods) or payment trail (for services) is accepted without
further proof, subject to rebuttable presumption of genuineness. In the
elevated tier, triggered by published indicators (a supplier registered
recently or with a thin filing history, a mismatch between supplier capacity
and invoiced value, a services supply in a category designated as high-risk,
abnormal patterns in the recipient's own purchases), the recipient is entitled
to a notice stating the indicator and may be required to produce a defined
dossier within a reasonable period. In the enhanced tier, triggered by
objective red flags (cancellation of the supplier's registration with
retrospective effect, adverse finding about the supplier in a completed
investigation), the Revenue may require the full evidentiary showing under
proposal 1. Safeguards: triggers published by notification; no tier may
be assigned by algorithmic score alone without recorded human review; tier
assignment reviewable.
3.
A statutory safe harbour with defined
limits. What: insert a provision protecting recipients
who have complied with the prescribed diligence. Why: to distinguish the
diligent from the complicit, consistent with Arise India and with the
separation of the three questions. How: where a supply is established
and the recipient has complied with the diligence prescribed for the relevant
tier at the time of the transaction, credit shall not be denied merely because
the supplier failed to pay tax, and the recovery shall be pursued against the
supplier first, with recourse to the recipient only on proof of knowledge or of
disregard of an indicator the recipient was notified of or ought to have
recognised. Where no supply is established, credit is denied and recovered with
interest, but penalty under the fraud limb, section 122(1)(vii) and prosecution
require proof of knowledge or recklessness. Safeguards: the safe harbour
should not operate where the recipient was a participant or
beneficiary-controller of the supplier; the burden to establish those
exceptions lies on the Revenue.
4.
Enhanced supplier authentication and
visible supplier-status signals. What: strengthen
verification at registration and give recipients a limited signal of supplier
status in the invoice-management interface. Why: the cheapest point at
which to stop a shell supplier is registration, and a recipient can use a
signal that the system already holds. How: the registration rules should
require physical-verification outcomes, bank-account linkage and, for high-risk
applications, biometric authentication, with state-level inspection records
feeding GSTN. The portal should show a limited-disclosure status (for example,
registration under verification, return-filing regularity) without disclosing
risk scores. Safeguards: a published appeal route against adverse
status; disclosure of status categories only; protection against
defamation-type harm to suppliers.
5.
Integrated verification across
invoice, e-way bill, payment and regulator data. What:
require systematic cross-matching and, for services regulated by other
statutes, structured data-sharing with the regulator. Why: the e-waste
matter shows that evidence of performance may sit in another regulator's
system. How: an inter-agency protocol under which designated GST
authorities may verify, in a defined and logged way, regulator-held records
(such as extended producer responsibility certificate generation and facility
throughput) for categories identified as high-risk. Safeguards: purpose
limitation, audit logs, notification to the taxpayer of the use made of the
data, and deference to the data-protection standard of proportionality.
6.
Digital-evidence standards for
fabrication allegations. What: prescribe by rule the
minimum disclosure when the Revenue alleges fabrication of electronic material.
Why: the weight of electronic evidence depends on method; fabrication
allegations carry penal consequences. How: disclosure of original files,
hash values, method statements and expert reports; reference to the notified
Examiner of Electronic Evidence where the allegation is contested; recording of
the standard of proof applied at each of the three layers in Part VIII. Safeguards:
right to inspect and to cross-examine, per Andaman Timber.
7.
Algorithmic alerts: human review,
reasons and audit. What: provide by statute or rule
that an automated risk output is not a ground for adverse action by itself. Why:
false positives, opacity, and the constitutional requirement of reasoned
decision-making. How: any rule 86A blocking, notice or ITC denial that
originates from an analytic alert must record the officer's independent
reasons; the taxpayer receives the category of indicator relied on; periodic
independent audit of model performance (false-positive rates by sector and by
taxpayer size) with an annual report to the GST Council. Safeguards: an
effective, time-limited review mechanism for blocking orders, taking account of
the Delhi High Court's treatment of rule 86A in Raghav Agarwal.
8.
Reasoned orders and a record of the
evidentiary basis. What: strengthen section 75(6) by
requiring an order denying credit for failure of proof to state the tier, the
evidence required, the evidence produced and the reason for rejecting it. Why:
to make the evidentiary standard reviewable on appeal. Safeguards:
prospective operation of any new standard; transitional protection for earlier
periods.
9.
A graduated penalty ladder. What:
confine the fraud limb and prosecution to knowing or reckless conduct and
prescribe a proportionate, lower civil consequence for non-compliance with
diligence standards where no knowledge is shown. Why: the present
structure allows a single shortcoming to be treated as both a failure of
entitlement and as fraud. How: amend sections 74A and 122 to require,
for the penal consequences, proof of knowledge or recklessness; ordinary
non-compliance attracts loss of credit with interest. Safeguards: the
Revenue bears the burden on mental state in adjudication, consistent with Hindustan
Steel.
C.
Objections and limits
Several
objections should be met. It will be said that tiered verification gives
fraudsters a map. The response is that published indicators can be broad, and
that the baseline tier is subject to post-claim audit; the framework regulates
the presumption, not the Revenue's power to investigate. It will be said
that safe harbours invite performative compliance. That risk is genuine: a
dossier can be manufactured. The answer lies in the independence criterion,
which relocates weight from the dossier to the third-party record, and in the
retention of the Revenue's power to rebut. It will be said that the framework
is expensive. It is, but the cost is borne where it can be borne most cheaply,
and the alternative is the status quo in which the cost falls unpredictably
after the event. Finally, the framework is not a cure for fraud that is truly
collusive and well funded; no verification regime is. Its purpose is to make
the law's demands specific, proportionate and answerable.
XII. Conclusion
Indian
ITC law is built on an assumption that documentary validity indicates
transactional authenticity. For the first generation of fraud the assumption
was tolerable; for fraud that fabricates the documents themselves, and
increasingly the evidence of performance, it is not. Ecom Gill
recognised the distinction, but its remedy of requiring further documents is
vulnerable to the same fabrication. The reported e-waste controversy, whatever
its eventual adjudication, illustrates the problem: services whose performance
the buyer cannot see, corroboration sourced from the supplier, and no
prescribed standard for what the buyer should have verified.
The
answer is neither to abandon documentary compliance nor to demand unlimited
verification. It is to determine, rule by rule, how far documentary compliance
suffices and when verification of the transaction becomes justified; to attach
evidentiary weight to independence of provenance; to keep entitlement, tax
receipt and culpability analytically separate; and to constrain verification
with published triggers, reasons, review and human judgment. The framework
proposed here is an attempt to protect, in order of the legal priorities the
problem demands, the revenue, the integrity of the credit chain, the honest
taxpayer, commercial certainty and procedural fairness. The first of these
cannot be secured at the cost of the last without the rule of law paying the
difference.
[1]Central Goods and
Services Tax Act 2017 (CGST Act), s 2(62), 2(63) and 16(1). The Act is
supplemented by the Integrated Goods and Services Tax Act 2017 (IGST Act) and
State GST Acts.
[2]'Fake ITC claims
detection by central GST officers up 51% at Rs 36,374 crore in FY24' Deccan
Herald (29 July 2024)
https://www.deccanherald.com/business/fake-itc-claims-detection-by-central-gst-officers-up-51-at-rs-36374-crore-in-fy24-3126661
accessed 5 October 2026.
[3]Saptarshi Datta,
Himanshu Thakkar and Siddharth Dabhade, 'Combating GST Input Tax Credit Fraud
in India: Forensic Accounting, Data Analytics, and Legal Enforcement
Strategies' (2026) 5(1) International Journal of Management, Public Policy and
Research 37, 42.
[4]Aditya Kalra,
'Exclusive-Haier, Godrej, Blue Star failed to detect fake e-waste recycling
claims, India notice shows' (Reuters, 30 September 2026), syndicated at
https://1027wbow.com/2026/09/30/exclusive-haier-godrej-blue-star-failed-to-detect-fake-e-waste-recycling-claims-india-notice-shows/
accessed 5 October 2026.
[7]CGST Act, s
16(2)(a)–(c). The conditions were restructured by the Finance Act 2021 and the
Finance Act 2022, with effect from 1 January 2022 and 1 October 2022
respectively, which inserted s 16(2)(aa) (supplier's furnishing of invoice
details in the outward statement, communicated to the recipient) and amended s
16(2)(c)..
[8]CGST Act, s 155.
[9]The CGST Act does
not define a 'fake invoice'. The term is used in a Standard Operating Procedure
circulated by the CBIC (Commissioner (Investigation)) to the GST Council
Secretariat, which describes fake invoices as GST invoices raised by an entity
without actual supply of goods or services or payment of GST: 'Standard
Operating Procedure (SOP) for Tackling Fake Invoice Cases'
https://gstcouncil.gov.in/sites/default/files/SOP-for-Tackling-Fake-Invoice.pdf
(as cited in the secondary literature; I have accessed the text as reproduced
at
https://taxguru.in/goods-and-service-tax/sop-utilising-fake-invoice-issuers-dataset-sgst-authorities.html
accessed 5 October 2026).
[10]CGST Act, ss 73
and 74 (and, for periods from 2024-25, s 74A) refer to input tax credit
'wrongly availed or utilised'. The distinction matters because blocking the
electronic credit ledger under CGST Rules 2017, r 86A operates on credit that
is available but not yet utilised.
[11]Michael Keen and
Stephen Smith, 'VAT Fraud and Evasion: What Do We Know and What Can Be Done?'
(2006) 59 National Tax Journal 861.
[12]Datta, Thakkar and
Dabhade (n 3).
[13]CGST Act, s 16(2)
and (4); CGST Rules 2017, r 36(1) (documents on which credit may be taken).
Section 16(5) and (6), inserted by the Finance (No 2) Act 2024.
[14]CGST Act, s 16(2),
Explanation.
[15]CGST Act, ss 37
and 41 as amended by the Finance Act 2022; GSTR-1 (outward supplies), GSTR-2B
(auto-generated statement of ITC) and GSTR-3B (summary return) are the
operative returns. The GST portal's Invoice Management System, introduced in
late 2024, lets recipients accept, reject or keep pending invoices before
GSTR-2B is generated: GSTN, 'Invoice Management System' advisory
https://tutorial.gst.gov.in/downloads/news/revised_advisory_on_ims.pdf accessed
5 October 2026.
[16]CGST Act, ss 35
and 36.
[17]CGST Act, s 31;
CGST Rules 2017, rr 46 and 48..
[18]CGST Rules 2017, r
138.
[19]CGST Act, ss 136
and 145.
[20]CGST Act, ss 73,
74 and 74A (inserted by the Finance (No 2) Act 2024 for financial years 2024-25
onwards). Under s 74, voluntary payment before notice attracts a penalty of
15%, within 30 days of notice 25%, within 30 days of the order 50%, failing
which the penalty is equal to the tax. Section 74A preserves a graded
structure, with sixty-day windows in the fraud limb.
[21]CGST Act, s
122(1)(ii), (vii) and (1A).
[22]CBIC, Circular No
171/03/2022-GST (6 July 2022), as summarised in 'Fake Invoice | Applicability
of Penalties & Demand | CBIC Clarification'
https://gsthero.com/blog/fake-invoice-applicability-of-penalties-demand-cbic-clarification/
accessed 5 October 2026.
[23]CGST Act, s
132(1)(b), (c), (1)(i)–(iv), (4) and (5).
[24]CGST Act, s 135.
[26]CGST Act, ss 67
and 83; CGST Rules 2017, r 86A. On the standard for r 86A and its relation to s
83, see Raghav Agarwal v Commissioner of Central Tax and GST WP(C)
10980/2024 (Delhi HC, 24 September 2024) (Vibhu Bakhru and Sachin Datta JJ).
[27]CGST Act, s 29(2);
CGST Rules 2017, rr 21 and 21A.
[28]CBIC, Instruction
No 02/2024-GST, F No CBIC-20/16/30/2020-GST (12 August 2024)
https://cbic-gst.gov.in/pdf/instruction-02-2024-GST-12082024.pdf.
[29]Marwin Heinemann
and Wojciech Stiller, 'Digitalization and Cross-Border Tax Fraud: Evidence from
E-Invoicing in Italy' (2025) 32 International Tax and Public Finance 195;
Matthieu Bellon and others, 'Digitalization to Improve Tax Compliance: Evidence
from VAT E-Invoicing in Peru' (2022) 210 Journal of Public Economics 104661.
[30]'GST revenues
remain high while fake invoicing menace continues' Deccan Herald
(opinion)
https://www.deccanherald.com/opinion/gst-revenues-remain-high-while-fake-invoicing-menace-continues-1228984.html
accessed 5 October 2026..
[31]CBIC Instruction
02/2024-GST (n 28).
[32]Kalra (n 4).
[33]ibid. Indian
outlets report the credit amounts in rupees at roughly Rs 3.53 crore (Haier),
Rs 1.79 crore (Blue Star) and Rs 1.53 crore (Godrej): 'Haier, Godrej and Blue
Star Face Scrutiny Over E-Waste Tax Credit Claims' The420.in
https://the420.in/haier-godrej-blue-star-e-waste-tax-credit-claims/ accessed 5
October 2026.
[34]Kalra (n 4).
[35]ibid.
[36]State of Karnataka
v Ecom Gill Coffee Trading Pvt Ltd Civil Appeal No 230 of 2023 and connected
appeals (SC, 13 March 2023) 2023 INSC 212 (MR Shah and CT Ravikumar JJ).
[43]Arise India Ltd v
Commissioner of Trade and Taxes, Delhi (Delhi HC, 26 October 2017), special
leave petition dismissed in Commissioner of Trade and Taxes, Delhi v Arise
India Ltd (SC, 10 January 2018). The Delhi High Court held that s 9(2)(g)
of the Delhi VAT Act, in so far as it denied credit to bona fide purchasers for
the seller's default, violated arts 14 and 19(1)(g) of the Constitution.
[44]Nitesh Jain,
'Supreme Court Judgement in case of Ecom Gill on ITC & Implications under
GST' TaxGuru (24 March 2023)
https://taxguru.in/goods-and-service-tax/supreme-court-judgement-case-ecom-gill-implications-gst.html
accessed 5 October 2026.
[45]Anil Rice Mill v
State of Uttar Pradesh Writ Tax No 886 of 2023 (Allahabad HC, 14 August
2024) (Piyush Agrawal J); see the account in PwC, 'Burden to prove ITC
eligibility under the GST law is on taxpayer' (Tax Insights, 20 August
2024)
https://www.pwc.in/assets/pdfs/news-alert/tax-insights/2024/pwc_tax_insights_20_august_2024_burden_to_prove_itc_eligibility_under_the_gst_law_is_on_taxpayer.pdf
accessed 5 October 2026..
[46]The Allahabad High
Court has also distinguished Ecom Gill on its facts where movement and
return compliance were established: see the account of Writ Tax No 3829 of 2025
(17 December 2025) in 'No Fraud, No Jurisdiction: Allahabad HC Draws a Clear
Line under GST Section 74' TaxGuru
https://taxguru.in/goods-and-service-tax/fraud-jurisdiction-allahabad-high-court-draws-clear-line-gst-section-74.html
accessed 5 October 2026..
[48]Bharatiya Sakshya
Adhiniyam 2023 (Act 47 of 2023), in force from 1 July 2024.
[49]BSA, ss 57, 61 and
63 and Schedule. The BSA replaced s 65B of the Indian Evidence Act 1872, whose
certificate requirement was authoritatively construed in Anvar PV v PK
Basheer (2014) 10 SCC 473 and Arjun Panditrao Khotkar v Kailash
Kushanrao Gorantyal (2020) 7 SCC 1. I rely on the secondary accounts of the
Schedule's form, e.g. 'Electronic evidence under the BSA, 2023' iPleaders
https://blog.ipleaders.in/electronic-evidence-under-the-bsa-2023/ accessed 5
October 2026.
[50]CGST Act, ss 136
and 145. I have not examined whether s 145 has been conformed to the BSA's
terminology.
[51]BSA, s 39 (expert
opinion, corresponding to s 45 of the Indian Evidence Act 1872); Information
Technology Act 2000, s 79A.
[55]CBIC Instruction
02/2024-GST (n 28).
[56]Datta, Thakkar and
Dabhade (n 3) 45; Madan and Sharma (n 3) 6–7..
[57]'Delhi
government's GST dept unearths Rs 347 crore "fake billing scam"' Deccan
Herald (PTI)
https://www.deccanherald.com/amp/story/india%2Fdelhi-governments-gst-dept-unearths-rs-347-crore-fake-billing-scam-1066953.html
accessed 5 October 2026. The report describes an official statement; the
allegations had not been adjudicated.
[59]Joined Cases
C-439/04 and C-440/04 Axel Kittel v Belgian State and Belgian State v
Recolta Recycling SPRL ECLI:EU:C:2006:446, paras 56–61.
[60]Council Directive
(EU) 2025/516 of 11 March 2025 amending Directive 2006/112/EC as regards VAT
rules for the digital age; Heinemann and Stiller (n 29).
[61]Modern Dental
College and Research Centre v State of Madhya Pradesh (2016) 7 SCC 353;
Justice KS Puttaswamy (Retd) v Union of India (Aadhaar) (2019) 1 SCC 1.
[62]CGST Act, s 168; Commissioner
of Central Excise v Ratan Melting and Wire Industries (2008) 13 SCC 1.
[63]On legitimate
expectation generally see Navjyoti Co-operative Group Housing Society v
Union of India (1992) 4 SCC 477.
[64]CGST Act, s 75(4)
and (6).
[65]Kranti Associates
Pvt Ltd v Masood Ahmed Khan (2010) 9 SCC 496; State of Uttar Pradesh v Sudhir
Kumar Singh (2021) 19 SCC 706.
[66]CGST Act, ss
107–112.
How to Cite This Article
ADITYA NIGAM, ANSH SHUKLA, From Documentary Compliance to Transaction Verification: A Critical Analysis of Fraudulent Input Tax Credit under the GST Regime., White Black Legal – International Law Journal, ISSN: 2581-8503, Vol. Volume 4, Issue VOLUME 4 ISSUE 2, October 2026, pp. 448-477. Available at: https://www.whiteblacklegal.co.in/details/from-documentary-compliance-to-transaction-verification-a-critical-analysis-of-fraudulent-input-tax-credit-under-the-gst-regime
Author & Publication Record
Authors: ADITYA
NIGAM & ANSH SHUKLA
Registration ID: 107157 | Published Paper ID: WBL7157 & WBL7158
Year: Oct- 2026 | Volume: 4 | Issue: 2
Approved ISSN: 2581-8503 | Country: Delhi, India
Page No.: 448-477
Full Text Preview
Open in New Tab
Copied