From Shubhi K | Product & Market Analysis

Vendor Email Compromise Is 61% of BEC. The Invoice Controls That Stop It

On this page

Vendor email compromise now makes up 61% of business email compromise, according to Abnormal AI's 2026 attack landscape report. The common story is that attackers hijack real supplier inboxes and send fake invoices from them. Abnormal's own data says that is true for only about 9% of these campaigns. The dangerous minority clusters in one request type, and that is where your accounts payable controls belong.

Key takeaways

  • Vendor email compromise is now the majority of business email compromise. Abnormal AI's 2026 report puts it at 61% of BEC, from nearly 800,000 attacks across 4,600+ organisations between July and December 2025.
  • Most vendor fraud emails are impersonation, not hijacked accounts. In Abnormal's labelled high-risk campaigns, 87.5% impersonated the vendor and 8.95% came from a genuinely compromised vendor account.
  • Compromised supplier accounts concentrate in bank detail change requests. Abnormal's pretext figures imply that about 70% of compromised-account vendor campaigns ask to update billing or bank details, by this post's arithmetic.
  • The control that works sits at the vendor master record, not the inbox. Nacha and the FBI both tell payers to confirm account changes through a channel already on file, never through contact details inside the request.
61%Share of business email compromise that is vendor email compromise. Source: Abnormal AI, April 2026.
26.5%Bank detail change requests sent from a genuinely compromised vendor account. Source: Abnormal AI, June 2026.
$3.05BBEC losses reported to the FBI in 2025, across 24,768 complaints. Source: FBI IC3 2025 report, April 2026.

Short answer. Vendor email compromise is a fraud where an attacker poses as one of your suppliers, usually to redirect a payment. The control that stops it is not better email filtering. It is a rule that no vendor bank detail change takes effect until someone confirms it by phone on a number already in your records, and a second person approves it.

What the 61% vendor email compromise figure measures

Abnormal AI is an email security vendor. Its 2026 attack landscape report draws on nearly 800,000 email attacks seen across more than 4,600 customer organisations between July and December 2025. The press release from 22 April 2026 states that vendor email compromise makes up 61% of all business email compromise.

Business email compromise, or BEC, is the broad category: an email that pretends to come from someone you trust and asks for money or data. Vendor email compromise, or VEC, is the subset where that trusted party is an external supplier rather than your own chief executive or finance team. In practice it means invoices, payment chases and bank detail changes.

BEC is a small share of attacks and a large share of losses

BEC is not the most common attack Abnormal sees. A July 2026 Abnormal post puts it at roughly 11% of attacks by volume, against 58% for phishing. Its weight comes from the money. The FBI logged 24,768 BEC complaints in 2025 and $3.05 billion in reported losses, second only to investment fraud.

Put those together and the shape is clear. A minority attack type carries an outsized share of the loss, and inside that type, supplier-themed messages are now the majority. If your team only trains people to spot a fake chief executive, it is training for the smaller half of the problem.

Every number from Abnormal comes from organisations that already pay for Abnormal. That is a self-selected population, and it skews toward firms large enough to buy dedicated email security. The figures are directional evidence about attack mix. They are not a census of all businesses, and I treat them that way throughout this post.

Most vendor email compromise is impersonation, not hijacking

The popular version of this threat is dramatic. An attacker breaks into a real supplier's mailbox, reads months of threads, then replies inside a genuine conversation with new bank details. That happens. It is not the typical case.

Abnormal's June 2026 breakdown of vendor email compromise looked at high-risk VEC campaigns with verified labels. Across all of them, 87.5% impersonated the vendor and 8.95% used a genuinely compromised vendor account. Impersonation here means a lookalike domain, a spoofed display name or a free webmail address dressed up as the supplier.

Four pretexts carry almost all of it

The same post splits those campaigns by the story the attacker tells. Invoice inquiries lead, followed by billing account updates, payment inquiries and requests for quote. Each has a very different compromise rate, and that difference is the most useful thing in the dataset.

What vendor email compromise campaigns ask for Share of high-risk VEC campaigns by pretext, with the share sent from a compromised vendor account. Invoice inquiry 41.3% Compromised account: 0.95% Billing account update 23.6% Compromised account: 26.5% Payment inquiry 19.8% Compromised account: 1.7% Request for quote 15.3% Compromised account: 12.8% Source: Abnormal AI VEC analysis, 3 June 2026. Customer data, July to December 2025.
Invoice inquiries are the most common pretext and almost never come from a real supplier account. Bank detail changes are the second most common, and more than a quarter of them do.

An impersonated vendor writes from an address that is not on file. A lookalike domain, a webmail account or a reply-to field pointing somewhere else can all be caught by a person or a filter that checks the sender against the vendor master record. Invoice inquiries are 98.1% impersonation, so a simple sender check removes most of the largest category.

That is the part of the problem your email security vendor is built for. The remainder is the part it is not, and that remainder decides your losses.

Bank detail change requests are where compromised supplier accounts concentrate

Abnormal's figures allow one calculation the post itself does not make. Multiply each pretext's share of campaigns by its compromise rate and you get the share of all VEC campaigns that are both that pretext and sent from a real supplier account. Add the four together and you land back on Abnormal's overall 8.95%, which is a useful check that the numbers are consistent.

Where compromised-account vendor campaigns sit, computed from Abnormal's pretext data
PretextShare of VEC campaignsCompromise rateShare of all VEC that is compromised and this pretextShare of compromised-account VEC
Billing account update23.6%26.5%6.25 pointsAbout 70%
Request for quote15.3%12.8%1.96 pointsAbout 22%
Invoice inquiry41.3%0.95%0.39 pointsAbout 4%
Payment inquiry19.8%1.7%0.34 pointsAbout 4%
Total100%8.95% reported8.94 points computed100%

Working: 23.6% x 26.5% = 6.25 points; 6.25 / 8.94 = 70%. Inputs from Abnormal AI's VEC analysis, 3 June 2026. The final column is this post's arithmetic, not an Abnormal figure, and assumes the published rates apply to the same campaign base.

Where the hijacked supplier accounts are used Share of compromised-account VEC campaigns, by pretext. Computed from Abnormal's published rates. Billing account update, about 70% Quote, 22% Bank detail changes are the request that redirects money already owed. Quote requests aim at sales teams. Invoice inquiry, about 4% Payment inquiry, about 4% Source: Zan Digital calculation from Abnormal AI VEC analysis, 3 June 2026. Share x compromise rate, normalised.
Seven in ten campaigns sent from a real supplier mailbox ask you to change where you pay. That one request type is the place to spend control effort.

Why attackers spend real accounts on bank changes

A compromised supplier mailbox is expensive to obtain and easy to lose once the supplier notices. Attackers appear to spend it where it pays most. A bank detail change does not need a fake invoice at all. It redirects money you already owe to a real supplier, on a real invoice, in an amount your team expects to pay.

That makes the change request the single highest-risk email your accounts payable team receives. It is also, in most firms I have seen described, handled by the same person who processes routine queries, often by email reply. I would treat any vendor bank change as a payment instruction in its own right, with the same approval weight as releasing the payment.

Abnormal reports that government agencies see billing account updates in 40.8% of their VEC campaigns, with a 41.2% compromise rate on that pretext. Nacha's guidance offers a plausible reason. Public-sector contracting details are often a matter of public record, so attackers know exactly which supplier to impersonate and when an invoice is due. Abnormal notes the government sample is relatively small.

Why sender checks miss the vendor email compromise that pays out

Email authentication standards such as SPF, DKIM and DMARC confirm that a message really came from the domain it claims. They are worth having, and they stop exact-domain spoofing. They cannot help when the message genuinely comes from the supplier's domain because the supplier's mailbox has been taken over.

That is the 26.5% case on bank detail changes. The domain is right. The thread may be real. The signature, tone and history are all the supplier's own, because the attacker is sitting inside the supplier's account. Every sender-based control passes, which is why these campaigns are the ones worth worrying about.

Older training told staff to look for poor grammar, odd phrasing and generic greetings. Fluent text is now cheap to produce with language models, so those tells are fading. The Abnormal pages I read publish no figure for the share of these attacks written with AI, so I will not invent one. The direction does not need a number to matter.

My view is that content inspection is the wrong layer to rely on for vendor fraud. If the control depends on a person noticing that an email feels off, it fails exactly when the attacker is good. Process controls do not care how convincing the email is. Our earlier analysis of why payment process beats deepfake detection reaches the same conclusion for voice and video.

Reply-based verification confirms the attacker

The most common weak control is replying to the email to ask the supplier to confirm. If the account is compromised, the attacker answers. Nacha's guidance is explicit on this point. Do not use reply when authenticating payment requests; forward to a known address instead. Calling a phone number printed in the same email has the same flaw.

The invoice fraud controls checklist for accounts payable

Each control below traces to published guidance from the FBI or Nacha, or to the Abnormal data above. I have ordered them by where they sit in the invoice and payment flow, from intake to release, because a control is only useful at the point where the fraud passes through.

Where each control sits in the payables flow Process map built from Nacha and FBI guidance. The red stage carries most compromised-account risk. 1. Intake sender check 2. Vendor master change 3. Invoice match 4. Dual approval 5. Daily review Catches lookalike and webmail senders Callback on number already on file PO and receipt before any ad hoc payment Two people for release and changes. Same-day check of payees against master. Sources: Nacha BEC and vendor impersonation guidance; FBI IC3 PSA I-091124-PSA, September 2024. Illustrative process map. It shows sequence, not measured effectiveness at each stage.
The second box is the one sender checks cannot reach. It is also where Abnormal's data puts most compromised supplier accounts, so it deserves the heaviest control.
Accounts payable controls against vendor email compromise, with sources
ControlWhat it stopsSource
Confirm any bank detail change using contact details already on file, never those in the requestCompromised-account and impersonated bank changesNacha BEC and VIF guidance; FBI PSA tip on secondary channels
Require two people to approve payment release and any change to recipient detailsA single employee being persuaded; internal fraudNacha BEC and VIF guidance
Forward, never reply, when checking a payment emailAttacker answering verification on a hijacked threadNacha BEC and VIF guidance
Check sender address against the vendor master; flag webmail and lookalike domainsImpersonation, which is 87.5% of high-risk VECFBI PSA tips; Abnormal VEC data
Share vendor payment forms only through secure means or with known entitiesAttackers obtaining blank change formsNacha BEC and VIF guidance
Review accounts daily and reconcile payees to the vendor masterLate discovery that puts recovery out of reachNacha BEC and VIF guidance
Treat a bank's call questioning a payment as a real alertIgnoring the one outside party that saw the anomalyNacha BEC and VIF guidance

The vendor master change is the control that matters most

Nacha's 2019 guidance on business email compromise and vendor impersonation fraud says to authenticate requests to change payment information using known contact information, "rather than contact information provided with the change request." Its example is a telephone number already on file. The FBI's September 2024 public service announcement gives the same instruction in its own words: use secondary channels or two-factor authentication to verify requests for changes in account information.

Write the rule so it has no exceptions for urgency or seniority. Urgency is a listed warning sign in Nacha's training guidance, alongside requests for secrecy. A supplier who is genuinely changing banks can survive a one-day delay. An attacker usually cannot, because the supplier will notice the missing payment.

Dual control on both changes and payments

Nacha recommends dual control, meaning two people approve both payment initiation and changes to payment information such as recipient details. The second person should be someone who did not handle the original email. That way a single persuaded employee cannot complete the fraud, and the same control blocks a class of internal fraud as a side effect.

Configure this in your ERP or payables system rather than in a policy document. A control that lives only in a handbook is skipped at quarter end. If your team is evaluating AI agents inside SAP and Oracle ERP, check that any agent able to edit vendor records is subject to the same second approval as a human.

Payment inquiries make up 19.8% of VEC campaigns in Abnormal's data, and Abnormal defines them as requests for ad hoc payments that may not reference an existing invoice or vendor relationship. A three-way match of purchase order, goods receipt and invoice is the standard answer. No match, no payment, whoever is asking.

Nacha encourages daily account review, and that control does not stop fraud. It shortens the time between payment and discovery. Our analysis of the BEC wire recall window shows why that matters: the FBI froze 58% of the money its recovery team pursued in 2025, and only fast reporters get into that process at all.

What the 2026 Nacha rules change for accounts payable teams

Nacha, the body that governs the ACH network, brought in new fraud monitoring rules in 2026. The Phase 1 rule page sets one duty for originators, third-party senders and banks. They must "establish and implement risk-based processes and procedures reasonably intended to identify ACH Entries initiated due to fraud" and review them at least annually.

Phase 1 took effect on 20 March 2026 for non-consumer originators with annual ACH origination volume of 6 million or more in 2023. Phase 2 removed that threshold. Nacha set it for 19 June 2026, a federal holiday, so the practical compliance date was 22 June 2026. If your company pays suppliers by ACH, you are an originator, and the rule now applies to you whatever your volume.

False pretenses names vendor impersonation directly

The rules introduce the term false pretenses. It means inducing a payment by misrepresenting your identity, your authority to act for someone else, or the ownership of the account to be credited. Nacha states that this covers business email compromise, vendor impersonation and payroll impersonation. It excludes disputes over fake or poor-quality goods.

The practical reading is that vendor payment diversion is now named inside the ACH rulebook. The rule does not prescribe callback verification, and it does not change who bears the loss. Nacha also says a risk-based approach cannot be used to conclude that no monitoring is needed. I would document the controls in the checklist above as your risk-based process, because an auditor or your bank will ask what it is.

Nacha's credit-push response checklist for corporate practitioners is useful after the fact. It tells you to determine whether the fraud came from a change request or an invoice, and whether it caused other payments. That second step matters for vendor fraud specifically, because a changed bank record will pay every future invoice to the attacker until someone corrects it.

Where this argument is weakest

The checklist above rests on two kinds of evidence. The controls come from FBI and Nacha guidance, which is primary and stable. The prioritisation comes from Abnormal's data, which is weaker, and that weakness should be stated plainly.

Every attack-mix figure here comes from Abnormal's own customers in the second half of 2025. Those customers chose to buy an email security product, and Abnormal sells protection against exactly this threat. The pretext and compromise rates also come from a labelled subset of high-risk campaigns whose size the blog does not publish. A different vendor's customer base could produce a different split.

The 70% figure is my arithmetic, not a measurement

The claim that about 70% of compromised-account VEC campaigns are bank detail changes comes from multiplying Abnormal's published shares by its published rates. The inputs reconcile to Abnormal's own 8.95% total, which is reassuring. It still assumes all four rates were measured on the same base, which the source does not confirm.

Campaign counts are not losses

Abnormal counts attack campaigns, not dollars lost. An impersonated invoice inquiry that succeeds might cost more than a compromised bank change that fails. No public dataset I found links VEC pretext to loss size, and the FBI does not split BEC losses by pretext. The case for prioritising bank changes rests on mechanism and frequency, not on measured loss share.

None of this weakens the controls themselves. Callback verification and dual approval are cheap and stop both impersonation and compromise. What the weaknesses limit is the precision of the ranking, not the direction.

Frequently asked questions

What is vendor email compromise?

Vendor email compromise is a form of business email compromise where an attacker poses as one of your suppliers. The aim is usually to redirect a payment by sending a fake invoice, chasing a payment or asking you to update the supplier's bank details. Abnormal AI's 2026 report found it makes up 61% of business email compromise across its customer base.

How is vendor email compromise different from business email compromise?

Business email compromise is the wider category covering any email that impersonates a trusted party to obtain money or data. Vendor email compromise is the subset where that party is an external supplier. Executive impersonation targets your own chain of command, while vendor fraud targets accounts payable through invoices, payment queries and bank detail changes that look like routine supplier admin.

How do you verify a vendor bank account change request?

Call the supplier on a phone number already in your vendor master record, never one printed in the request. Nacha guidance says to use known contact information rather than details supplied with the change. Then have a second person approve the change in your payables system before any payment uses it. Do not reply to the original email to confirm.

Does DMARC stop vendor email compromise?

Only partly. DMARC, alongside SPF and DKIM, stops attackers sending email that falsely claims to be from an exact domain. It does not stop lookalike domains, and it cannot help when a real supplier mailbox has been taken over. Abnormal found 26.5% of bank detail change attacks came from compromised vendor accounts, which pass every sender check.

What invoice fraud controls should accounts payable have?

The core set is callback verification of any bank detail change using contact details on file, plus dual approval of payment release and vendor record changes. Teams should also forward rather than reply when checking payment emails, and match purchase orders before ad hoc payments. Add daily review of outgoing payments against the vendor master. Nacha and the FBI recommend all of these.

Do the 2026 Nacha rules apply to companies paying vendors by ACH?

Yes, for non-consumer originators. Phase 1 took effect on 20 March 2026 for originators with 6 million or more ACH originations in 2023. Phase 2 removed the volume threshold, with a practical compliance date of 22 June 2026. Originators need risk-based processes to identify ACH entries initiated by fraud, including vendor impersonation.

Where to start

Open your payables system today and find out who can change a vendor's bank details, and whether that change needs a second approver. If one person can do it alone, that is the gap the data points at, and it is usually a configuration change rather than a project.

Then pull the last 90 days of vendor bank changes and check each one for a recorded callback to a number that predates the request. If the record does not exist, assume the control does not either. The piece on what SOC 2 auditors accept as evidence is a useful model for how to log it from here.

Read alongside

This post covers stopping the payment. For what to do in the first hour after one leaves, read the BEC wire recall window, measured. For the wider shift in fraud tooling, see the analysis of AI-enabled cybercrime growth.

References

  1. Abnormal AI, 2026 Attack Landscape Report press release, 22 April 2026. Used for the 61% share and the sample description.
  2. Abnormal AI, vendor email compromise analysis from the 2026 Attack Landscape Report, 3 June 2026. Used for pretext shares, impersonation and compromise rates.
  3. Abnormal AI, 2026 attack landscape: business as usual risk, 23 July 2026. Used for BEC and phishing shares of attack volume.
  4. FBI Internet Crime Complaint Center, 2025 Internet Crime Report, April 2026. Used for BEC complaints and losses.
  5. FBI Internet Crime Complaint Center, Business Email Compromise: The $55 Billion Scam, PSA I-091124-PSA, 11 September 2024. Used for FBI prevention guidance.
  6. Nacha, Business Email Compromise, Vendor Impersonation Fraud, and Payments, April 2019. Used for callback, dual control, forward-not-reply and daily review guidance.
  7. Nacha, Risk Management Topics: Fraud Monitoring Phase 1, rule page retrieved November 2026. Used for dates, thresholds and the false pretenses definition.
  8. Nacha Risk Management Advisory Group, Responding to Credit-Push Payments Fraud: A Checklist for Corporate Practitioners, 2024. Used for post-incident steps.

The weakest part of this source base is that every attack-mix figure comes from a single security vendor's customer data, with the labelled subset's size unpublished. The FBI and Nacha sources are primary but publish guidance, not effectiveness data for individual controls. Figures last verified 1 November 2026.

RR
Ritu Raj
Founding Member at Zan Digital. Writes about AI product economics, B2B software markets and the operating controls that sit underneath vendor claims.

Related reading