Business email compromise: how invoice fraud works

How a fake change of bank details reaches the accountant, what traces it leaves in the email thread, and which payment controls stop BEC fraud.

Topic
IT security
Author
ARLITECH expert team
Published

Invoice fraud usually doesn't even need malware, so the antivirus has nothing to find. An accountant gets an email from a known supplier saying the company has switched banks and the next invoice should be paid to the new account. The accountant updates the account number, the money goes out, and two weeks later the real supplier sends a payment reminder.

This is business email compromise, or BEC. The attack runs through email and through the company's payment process, so the defense has to cover both: the mail system's settings and the way the company approves transfers. This article walks through how such a fraud is put together, what technical traces it leaves, and what to do if the money is already gone.

What BEC is and how it differs from mass phishing

Mass phishing sends the same email to thousands of addresses and relies on a small share of recipients clicking a link or entering their password. BEC is targeted. The attacker picks a specific company and finds out who pays the invoices, who the suppliers are, who runs the company, and how these people usually write to each other. The email fits into a real business situation and is often the continuation of a genuine thread.

A BEC email rarely contains a link or an attachment. The request itself is the attack: send the money here, change the account number, buy gift cards, forward the payroll register. That is why a traditional mail filter that looks for malicious files and links often finds nothing wrong with it. These emails rely on social engineering and trade on the sender's authority.

The scale of the damage shows in the report on 2024 from the FBI's Internet Crime Complaint Center (IC3), published in April 2025. It puts reported losses linked to BEC in 2024 at about $2.77 billion, the second largest of the report's categories. The figure only covers cases reported to IC3, so the real damage is higher.

Spoofed and compromised mailboxes

An attacker has two ways to send an email that appears to come from a known partner or from your own management.

Spoofed sender

The simplest is a spoofed display name. The mail client often shows only the name, for example the CEO's, while the address behind it is a free webmail account. On a phone, where the client shows even less, this gets through especially easily.

The second method is a lookalike domain. The attacker registers a domain that looks like the partner's at a glance: one letter swapped, a hyphen added, .com instead of .hu, or the letter pair “rn” standing in for “m”. The attacker can also set up their own SPF, DKIM, and DMARC records for this domain, so the email passes every authentication check, because technically it really does come from that domain.

The third is spoofing the real domain. Properly configured SPF, DKIM, and DMARC filter this out, provided the receiving side enforces the DMARC policy. Where the sender's domain has no strict DMARC policy, the spoofed email can still get through. The article on mail domain authentication covers the three records in detail.

Compromised mailbox

It is more dangerous when the attacker takes over a real mailbox: the supplier's accountant's, a colleague's in your own finance team, or an executive's. This usually takes a password or session token stolen with an earlier phishing email. An email sent from a compromised account passes every technical check, arrives with the usual signature and the earlier messages quoted, and the recipient sees nothing unusual in its form.

In a compromised account the attacker often just reads for weeks. They look at which invoices are due, who writes to whom and in what tone, and step in at the best moment, for example right after a large invoice has gone out. If the supplier's account was taken over, the fraud hits the buyer, even though nothing went wrong in the buyer's own systems.

Typical scenarios: changed bank details and the urgent executive request

Changed bank details

In one common scenario the attacker, posing as a supplier, says the bank account number has changed. The reason sounds plausible: a new bank, an audit, a reorganization, an account held by the parent company. Sometimes they also send a modified PDF invoice on which only the account number differs. The email often lands in a real billing thread, and the attacker handles the replies too, so that a suspicious follow-up question never reaches the real partner.

A variant targets payroll: posing as an employee, the attacker asks for the salary to be paid into a different account from now on.

The urgent executive request

In the other classic version the email comes from the CEO, or appears to. It concerns a confidential deal, an acquisition or an advance payment to a foreign partner, and asks for an immediate transfer, often at the end of the working day or on a Friday afternoon. The email says the executive can't be called right now because they are in a meeting or on a plane, and that nobody must be told about the matter. The same pattern appears when the request is to buy gift cards, or when the email forwards a payment request from an outside party posing as a lawyer.

What the two scenarios have in common is that the email asks the recipient to skip the usual checks, by rushing them, by demanding secrecy, or by claiming the usual contact is out of reach.

If an email asks for a financial change, pushes for speed, and asks you not to check it the usual way, all at once, confirm the request through a channel independent of the email, whoever the sender is.

Technical signs: lookalike domain, reply-to address, mailbox rule

A BEC email or a compromised account leaves traces in several places. Some are visible to the recipient, while others only the IT administrators will find in the logs.

The sender address and the lookalike domain

The first step is always to check the full sender address, not only the display name. A lookalike domain slips past a quick read, so it helps if the mail system flags a newly registered domain or a sender writing to the company for the first time. Many mail filters let you set up impersonation protection for executives' names and key partners' domains, along with a warning banner on email from outside the organization.

The reply-to address

In the email header the From and Reply-To fields can differ. The attacker can leave the sender as the real partner's address and point the reply-to address at their own mailbox. If the accountant answers with the “Reply” button, the message goes to the attacker. Where Reply-To points to a domain other than the sender's, a warning from the mail filter or the client can flag it clearly.

The mailbox rule

In a compromised mailbox the attacker often creates a rule. One kind forwards incoming email to an external address, so the attacker can keep reading it even after a password change. The other kind deletes the partner's replies or messages containing certain keywords, such as “invoice”, “transfer” or “bank account”, or moves them to a rarely checked folder such as RSS Feeds. The MITRE ATT&CK knowledge base lists both as separate techniques.

These rules are worth blocking and alerting on at tenant level: allow automatic forwarding to external addresses only for named exceptions, and set up an alert for every new rule that deletes messages or forwards them to an external address. Where all this is configured in a cloud mail tenant is described in the article on tenant security settings.

Sign-in traces

A compromised account also shows up in the sign-in log: a sign-in from an unusual country or provider network, a new device, a newly registered MFA method, or a sign-in that came right after a phishing email was opened. These signs are only useful if someone watches the log and the logs are kept long enough.

Process controls: callback on a known number, dual approval

When an email gets past every filter, the payment process is the next line of defense. The goal is that no payment detail can change on the strength of a single email, even if the accountant doesn't recognize the fraud.

Callback on a known number

Every change of bank account number has to be confirmed by phone, on a known number recorded earlier. That number should come from the contract, the supplier master data or an older email, not from the email asking for the change, because in that one the attacker gives their own number. A callback is just as necessary for an executive's payment request, even if the email says the executive is unavailable.

Dual approval

A change of bank account number in the supplier master data, and the first payment to a new payee, are approved by two different people. That way a single deceived employee can't push the change through. Banking systems can also be set to require two signatories above an amount threshold or for a new payee.

Further rules

  • A waiting period after a change: the first payment to the new account number goes out only after the callback and after a few days have passed.
  • A written rule that partners know about too: the company never asks for an account number change by email and never asks for a secret, urgent transfer.
  • Payee verification: under Regulation (EU) 2024/886 on instant credit transfers, for euro transfers, instant and standard alike, the bank has to check the payee's name against the account number before the transfer. In the euro area this has been mandatory since October 9, 2025, and in member states outside the euro area, Hungary included, it applies from July 9, 2027. The regulation does not cover forint transfers. Where the bank runs such a check and reports a mismatch, take it seriously.
  • Training for finance and assistant roles: staff should know that the callback is a mandatory step, and that sticking to the check when an executive is pushing is not a mistake.

A common weak point is an exception in the process: an executive whose requests nobody usually checks, or a partner so familiar that a callback seems rude. The rule has to cover these cases as well.

What to do if the transfer has already gone out

Here time matters most. With an instant transfer the money is in the other account within seconds, and fraudsters typically move it on or withdraw it quickly. The sooner the company acts, the better the chance that the receiving bank can still freeze the amount.

  1. Call your own bank right away rather than reporting it by email. Ask for a recall of the transaction and for the bank to contact the receiving bank. The bank will tell you what statement you need to sign.
  2. File a police report. Banks often make further steps conditional on it, and the investigation needs it too.
  3. Notify the real partner on a known phone number. If their mailbox was taken over, they have to act as well, and their other customers may be at risk.
  4. Investigate your own email. If the email came from the company's own account or in the name of a colleague, change the password, revoke active sessions, remove the MFA methods the attacker registered (or all of them if that is unclear, and have the user register again), review mailbox rules and forwarding, and save the sign-in and audit logs before they are overwritten.
  5. Preserve the evidence: the original emails with full headers, the modified invoice, the payment confirmation, and your correspondence with the bank.

If the investigation shows that a mailbox really was compromised, that may also count as a personal data breach, because the attacker could read the account's email. Management and the data protection officer need to know about it. The general steps for the first hours are covered in the article on the first 24 hours after an incident.

Related service: Email security →

Contact

Ready to secure your critical systems?

Our experts will assess where your security stands and build a roadmap that fits your organization.

ARLITECH's office in Gödöllő

Visit us

You are welcome to visit our Gödöllő office in person.

Address
Centrum Office Building, 2100 Gödöllő, Dózsa György út 13, 3rd floor, office 309
Office hours
Mon–Fri, 9 a.m. to 5 p.m.
Get directions (opens in a new tab)