← All posts

Business Email Compromise: How a Forged CFO Email Becomes a $187K Wire Transfer

BEC is the costliest cybercrime in the FBI's statistics — and it starts on a Windows endpoint. What your machine sees before the wire request ever arrives.

The wire transfer looked routine. The email came from the CFO's real address, referenced a real vendor, used the right tone, even quoted the correct invoice number. The finance officer sent $187,000 to a new account "per the CFO's updated banking details." The money was gone within the hour.

That is business email compromise (BEC), and it is the most expensive cybercrime in the FBI's Internet Crime Complaint Center statistics year after year — not because the technology is exotic, but because it isn't. BEC rarely starts with a zero-day. It starts with an endpoint.

How the attack actually works

A BEC campaign almost always follows the same four stages, and three of them happen on a Windows machine:

  1. Initial access. A phishing email delivers a credential harvester, or a stealer drops on the machine and exfiltrates browser-stored passwords and session cookies. The attacker now has a mailbox password — or, better, a live session token that bypasses MFA entirely.
  2. Silent observation. The attacker reads the mailbox for weeks. They learn the CFO's writing style, the approval chain, the vendor names, the invoice cadence. They set a mail rule that silently forwards or deletes messages so the victim never sees the reply that would expose them.
  3. Impersonation. From the real account, the attacker emails finance: urgent payment, new bank details, "I'm in a meeting, handle this now." The urgency is the payload — it suppresses verification.
  4. Cash-out. The money moves to a mule account and is layered away before anyone reconciles.

Notice what the victim's endpoint could have caught at stage one and two, long before the wire request ever arrived.

What your endpoint sees first

BEC is a people problem with machine symptoms. WinSentinel audits the machine-side conditions that make stages one and two possible:

None of these is a BEC detector by itself. Together, they describe whether a machine is a soft target for the exact intrusion BEC depends on.

The defense that actually works

Three controls, in order of leverage:

  1. Kill the credential path. Move to passwordless sign-in (Windows Hello for Business) where you can, and audit stored credentials everywhere else. A password that doesn't exist can't be phished, and a session that expires quickly is worth less to a thief.
  2. Verify payments out-of-band, always. Any change of banking details gets a phone call to a known number — never the number in the email. This is a process control, and it is the only one that stops a perfect impersonation. Write it down, train it, and make skipping it a violation of procedure rather than a judgment call made under time pressure.
  3. Audit the endpoint before the incident. Run WinSentinel's full audit — all 64 modules, free on a single machine — and fix the credential, browser, mail-rule, and logging findings first. BEC attackers pick the softest mailbox in the org; the audit tells you which one that is before they do.

The email will always look real. That is the point of the attack. The question is never "would I spot the fake" — it is whether the attacker could get into the mailbox in the first place, and whether anything on that machine would have noticed them doing it.

When a fleet is in play, the same findings matter at scale: one compromised mailbox is an incident, but a fleet-wide view shows whether the credential and mail-rule weaknesses are a single machine or a pattern across the org — that centralized picture is what WinSentinel Pro is for. On a single machine, though, the audit alone tells you the truth: either the door is open, or it isn't. Check before the CFO's name shows up in your inbox asking for money.