Americans filed 191,561 phishing complaints with the FBI in 2025 and lost about $216 million to them. They filed 24,768 complaints about business email compromise — roughly an eighth as many — and lost $3.05 billion. Fourteen times the money from a fraction of the reports.
That ratio is the most useful thing anyone can tell you about BEC. It isn’t a clever technical attack, it doesn’t involve malware, and in most cases there’s no malicious link or attachment anywhere in sight. It’s a business process attacked through a real email account by someone who has been reading quietly for weeks. This guide explains how it actually works from the attacker’s side, the forms it takes, the single assumption that lets it succeed, what to do in the first hour if money has already moved — which matters far more than most people realise — and the one control that genuinely prevents it.
2. Why it’s the most expensive crime in the data
3. The assumption that makes it work
4. How a BEC attack actually unfolds
5. The mailbox rule that hides everything
6. The forms it takes
7. If money already moved: the first hour
8. What actually prevents it
9. Frequently asked questions
1. What business email compromise actually is
Business email compromise is fraud carried out through email that a victim has every reason to trust, in order to redirect a payment or extract sensitive information. The defining feature is the trust: either the message comes from a genuine, compromised mailbox, or it’s crafted to sit so convincingly inside an existing relationship that nobody questions it.
That makes it a different species from ordinary phishing. A phishing email casts wide and hopes someone clicks. BEC is targeted, patient, and frequently contains nothing a security filter can object to — no link, no attachment, no spoofed sender. Just a short message about a bank account, sent at a moment when a payment was expected anyway.
The term covers several scenarios, but they share one shape: a person with authority to move money or release data receives a request that appears to come from someone entitled to make it. Everything else is detail.
2. Why it’s the most expensive crime in the data
The $3.05 billion figure needs context, because it’s reported losses from complaints actually filed with one agency in one country. The real total is certainly higher, since many organisations never report, particularly smaller ones that absorb the loss quietly.
What makes BEC so costly per incident is that it doesn’t steal small amounts from many people. It redirects single payments that were already going to happen, at whatever size that business normally transacts. A firm that routinely wires $80,000 to a supplier loses $80,000, not the $200 a consumer scam takes. There’s no technical ceiling on the amount, only the size of the invoice the attacker found while reading.
Most cybercrime steals something and then has to turn it into money. BEC skips that step entirely. The payment was already scheduled, already approved, and already expected — the attacker only changes where it lands.
3. The assumption that makes it work
Almost everyone believes they’d recognise a fraudulent payment request. It would feel off. The English would be stilted, the timing odd, the sender address subtly wrong, the tone not quite like the person it claims to be.
That belief is reasonable and it is the exact thing being exploited. By the time the fraudulent message arrives, the attacker has typically been inside the mailbox for weeks. They know the tone because they’ve read hundreds of examples of it. They know the timing because they’ve watched the invoice cycle. And the sender address isn’t wrong, because it’s the real address — they’re sending from the account itself.
There’s a quieter second assumption underneath, which is that a compromised email account would announce itself through spam, lockouts or alerts. It usually doesn’t, for a reason covered in the next section.
4. How a BEC attack actually unfolds
Researchers at UC Berkeley and Columbia, working with a security firm’s data, analysed nearly a million audit log events from 989 compromised enterprise accounts and then examined a random sample of 159 in detail. What they found describes this attack better than any list of warning signs.
Access. Usually a phishing page that relays the login to the real service in real time, capturing both the password and the session that’s issued afterwards. That technique defeats most forms of multi-factor authentication, not by breaking it but by waiting for the victim to pass it and stealing the result. Stolen credentials bought from elsewhere are the other common route.
Reading. This is the part that surprises people. In the study, 98% of attackers accessed email applications and 78% accessed nothing else at all — no files, no cloud storage, no attempt to move elsewhere. Activity lasted at least a day in 51% of accounts and at least a week in 37%. And in only 2 of the 159 accounts was the password changed during the intrusion. Locking the owner out would have been trivial; it would also have ended the operation. Among accounts that eventually sent anything malicious, 63% waited more than three days after logging in.
What’s being gathered during that time isn’t data in any sense a security tool would recognise. It’s who approves payments and who merely forwards them, which suppliers invoice monthly and on what date, how a particular colleague writes when they’re rushed, who’s on leave next week, and the exact format of an invoice that’s been paid without comment a dozen times. None of it is confidential. Together it’s worth more than any single document in the account.
The ask. When it comes, it usually isn’t a forged email at all. A real invoice arrives in a real thread from a real supplier, with several messages of ordinary conversation above it, and the attacker replies from the mailbox they’ve been living in. Same signature, same tone, same thread, plus a sentence explaining that the bank details have changed. There’s no spoofed address to inspect, no link to hover over and no attachment to scan — which is precisely why the usual defences find nothing to object to.
5. The mailbox rule that hides everything
One housekeeping step keeps the whole operation invisible, and it’s worth knowing in detail because checking for it takes thirty seconds.
In a case published this year by an incident response firm, an intruder created a mailbox rule that matched any sender address containing the “@” character. That matches every email that arrives. The rule moved all of it to another folder and marked it as read. The legitimate owner simply stopped seeing their own incoming mail.
The purpose isn’t theft. It’s control over what the owner learns and when. When colleagues began replying to ask whether that strange message had really come from them, those replies were swept away too. The person whose account it was had no idea anything was wrong, because the channel that would have told them was the channel the attacker had taken.
Variations show up constantly: rules that move anything containing the word “invoice” or a supplier’s name, rules that forward copies to an outside address, rules that delete bounce notifications. They tend to have blank or single-character names, because a rule called “.” attracts less attention than one called “hide”.
6. The forms it takes
Supplier invoice fraud. The most lucrative version. The compromised mailbox often belongs to the supplier rather than the victim, which means the target’s own security was never involved. A long-standing vendor appears to update their bank details, and the next scheduled payment goes to the attacker.
Executive impersonation. A message that appears to come from a chief executive or finance director, asking someone junior for an urgent transfer, usually with a reason for secrecy attached — an acquisition, a confidential deal. It exploits hierarchy rather than relationship, and works because people are reluctant to interrogate their boss.
Payroll diversion. An email to HR or payroll, apparently from an employee, asking to update the bank account for salary payments. Lower value per incident and often slower to detect, because nobody notices until payday.
Data requests. Not all BEC chases money directly. A request from “HR” or “finance” for employee tax records or payroll summaries yields material that funds identity fraud later.
Increasingly, these are reinforced by a phone call. If a payment request is followed by a voice that sounds like the person who sent it, the remaining doubt tends to evaporate — and synthetic voice has made that cheap to produce, as our guide on whether deepfakes are a real security threat explains.
7. If money already moved: the first hour
Here is the part most articles leave out, and it’s the most valuable thing on this page. People assume money sent by wire fraud is gone. Often it isn’t, if you move fast enough.
In 2025 the FBI’s Recovery Asset Team initiated its Financial Fraud Kill Chain on 3,900 incidents involving roughly $1.16 billion in attempted theft, and froze $679 million of it — a 58% success rate. That process works by contacting the receiving bank fast enough to stop the money being moved on or withdrawn, which means its entire effectiveness depends on how quickly someone picks up the phone.
Call your bank first, not your IT team. Ask them to recall or reverse the payment immediately, and request a Hold Harmless Letter or Letter of Indemnity. This is the step with a closing window, so it comes before everything else, including working out what happened.
File at ic3.gov straight away. Include every piece of banking information you have — originating and receiving account details, amounts, dates, reference numbers. The FBI’s guidance is explicit that complaints must contain all the required data for the kill chain to be actionable, and equally explicit that you should file regardless of the amount lost. If you’re outside the US, report to your national fraud reporting body with the same urgency.
Then contain the account. Change the password, sign out of all sessions — a stolen session can outlive a password change — and check mailbox rules, forwarding addresses and recovery details for anything you don’t recognise. Our guide to what actually happens during a security incident covers the rest of the response.
Warn the other side. If the compromised mailbox was your supplier’s, their other customers are being targeted with the same thread. If it was yours, your customers are. Tell them by phone, not by the email account in question.
8. What actually prevents it
Once you accept that an attacker can achieve perfect context — the right tone, thread, timing, format and relationship — a conclusion follows that people find unsatisfying because it isn’t clever.
No change to payment details is ever actioned on email alone. Confirm by voice, on a number you held before the conversation started. Not the number in the signature, not the number on the invoice, not a number supplied in the thread. This single rule defeats the entire category, because it’s the one step that happens outside the channel the attacker controls. Everything else on this list is secondary to it.
Give junior staff explicit permission to delay a payment. Most BEC succeeds in the gap between someone feeling uneasy and feeling allowed to say so. A written rule that nobody will ever be criticised for holding a transfer to verify it is worth more than most security software.
Check mailbox rules regularly. Make it a habit for anyone handling payments, and part of your response any time an account is suspected of compromise. It’s where this intrusion hides, and almost nobody looks.
Use phishing-resistant sign-in where you can. Passkeys and hardware security keys are bound to the genuine site’s address, so the relay attack that starts most of these fails outright. Our guide on multi-factor authentication explains why the type matters so much here.
Store verified contact details outside email. A supplier’s phone number kept only inside the email thread is a phone number the attacker can edit.
Train on the relationship, not the red flags. Advice built on bad grammar and odd sender addresses actively misleads people here, because a genuine BEC message has neither. The thing to be suspicious of is the request — any change to where money goes — regardless of how perfect the message looks. Our guide on social engineering covers why urgency and authority work so reliably on careful people.
9. Frequently asked questions
What is business email compromise in simple terms?
It’s fraud carried out through a trusted email account, usually to redirect a payment. An attacker gets into a real mailbox, reads it for days or weeks to learn how the business works, then sends a convincing request to change bank details. Victims reported $3.05 billion in BEC losses to the FBI in 2025.
How is BEC different from phishing?
Phishing is sent broadly and relies on someone clicking a malicious link. BEC is targeted and often contains no link, no attachment and no spoofed address, because it’s sent from a genuine compromised account into an existing conversation. That’s why email filters rarely catch it.
Can you get your money back after a BEC attack?
Often, if you act within hours. In 2025 the FBI’s Recovery Asset Team froze $679 million of roughly $1.16 billion in attempted theft across 3,900 incidents, a 58% success rate. Call your bank immediately to request a recall, then file at ic3.gov with full banking details. The FBI advises reporting regardless of the amount.
How do I know if my email account has been compromised?
Check your mailbox rules and forwarding settings for anything you didn’t create, especially rules that move or delete incoming mail, or forward to an outside address. Also review recent sign-in activity. Attackers usually avoid changing your password, so everything can look normal while they read.
Does multi-factor authentication stop business email compromise?
It helps a great deal but isn’t sufficient alone. Many BEC intrusions start with a phishing page that relays your login to the real site and steals the session issued after you pass MFA. Phishing-resistant methods such as passkeys or hardware security keys resist this, because they’re tied to the genuine website’s address.
Security, Decoded.
If this changed how you think about email fraud, you’ll like the newsletter. One security idea, decoded clearly, every week, free.
Subscribe to SKB Decoded →→ What Is Phishing? How to Spot It and What to Do
→ What Is Multi-Factor Authentication? Not All MFA Is Equal
→ What Is Social Engineering? (And Why Smart People Fall For It)





