SMSMC Blog

A compromised Microsoft 365 mailbox: the first 24 hours

What to do, in order, when a staff mailbox in Microsoft 365 has been taken over: contain the account, remove what the attacker left behind, check for payment fraud, then investigate.

By Jim Venuto

The call that starts it

A customer phones the office manager to ask why the company’s bank details changed. The email came from the bookkeeper’s real address, in the middle of a real thread about an open invoice, and it asked that the next payment go to a new account. The bookkeeper never sent it. Her mailbox has been taken over, and the person who took it over has been reading her mail for some time.

This is business email compromise (BEC), and it is not rare. The FBI’s Internet Crime Complaint Center (IC3) received 24,768 BEC complaints in 2025 with reported losses of just over $3.04 billion, the second-largest loss category in its 2025 report after investment fraud.

The first day decides most of the outcome. Money can sometimes be frozen if the bank hears quickly, and an attacker who keeps a foothold will come back. Below is an order of steps built on Microsoft’s own response guidance for Microsoft 365.

Hour 0 to 1: stop the bleeding

Call the bank if money moved

If any payment went out on the strength of a fraudulent email, this comes before anything technical. The IC3 2025 report is direct about it: time is of the essence. Contact your financial institution immediately, ask for a recall of the funds and any indemnification documents they need, and then file a complaint at ic3.gov with the full transaction details, whatever the amount. The IC3’s Recovery Asset Team works with banks to freeze funds; in 2025 it reported freezing about $679 million of roughly $1.16 billion in attempted theft across 3,900 incidents. That only works when the report is fast.

If a customer or supplier was the one who paid, tell them the same day so their bank can act.

Disable the account, then revoke sessions

Microsoft’s guidance for a compromised mailbox puts these first:

  1. Disable the user account during the investigation. Microsoft calls this the preferred step. If you cannot disable it, reset the password to a strong, unique one, and do not send the new password to the user by email, because the attacker may still be reading the mailbox. If the account syncs from on-premises Active Directory, reset it there, twice.
  2. Revoke the user’s sessions. A password reset alone does not end sign-ins that already happened. In the Microsoft Entra admin center, the user’s Overview page has a Revoke sessions button. Microsoft notes that access tokens last an hour by default, so an existing session can keep working for up to an hour after revocation; watch the sign-in logs during that window.
  3. Remove app passwords if any exist. Microsoft points out that app passwords are not revoked when the password is reset.

Do this from a computer you trust, signed in with an admin account that is not the compromised one.

Hour 1 to 4: remove what the attacker left behind

Attackers who get into a mailbox set things up so they can keep watching after you change the password. Microsoft lists the places to check. Go through each one.

CheckWhat to look forWhere
Sign-in methodsAn authenticator app, phone number or device you do not recognize, added so the attacker can pass MFAEntra admin center, user’s authentication methods
Inbox rulesRules that forward mail outside the company, or move messages to Notes, Junk Email or RSS Subscriptions so the user never sees repliesExchange Online PowerShell (Get-InboxRule -IncludeHidden); hidden rules do not appear in Outlook
Mailbox forwardingA forwarding address set on the mailbox itself, separate from inbox rulesExchange admin center or PowerShell
Consented appsThird-party apps the user granted access to mail or filesEntra admin center, enterprise applications
Admin rolesAny role assigned to the user that should not be thereMicrosoft Entra, Azure, Microsoft Purview and Defender portal role assignments
Contact detailsChanges to the user’s name, phone or address in the directoryMicrosoft 365 admin center

The inbox rule is the one most small businesses miss. A rule that quietly moves any message containing “invoice” or “payment” into a folder the user never opens lets the attacker hold a conversation with your customer without the bookkeeper noticing. Microsoft’s guidance specifically says to include hidden rules when you list them.

Consented apps matter too. If the user was tricked into approving an app, that app may hold its own access that survives a password change. Remove anything that should not be there.

Hour 4 to 12: find out what happened

With the account contained, work out how far this went. Microsoft recommends reviewing logs from just before the first suspicious activity through to the end of remediation:

  • Entra sign-in logs: IP addresses, locations, times and whether each sign-in succeeded. Look for the first sign-in you cannot explain.
  • The audit log in the Microsoft Defender portal (also in Microsoft Purview): filter by date range starting just before the suspicious activity, and do not filter by activity type on the first pass. You are looking for rule creation, forwarding changes, mailbox access and file downloads.
  • Message trace and the user’s Sent Items: what the attacker sent, and to whom.

Check how long your logs go back before you need them. Microsoft states that Audit (Standard) records are kept for 180 days by default, and that Audit (Premium) keeps Exchange, SharePoint, OneDrive and Entra records for one year for users with E5-level licensing. If the attacker was in for longer than your retention, the earliest evidence is gone, so export what you have now.

Write down a timeline as you go: first suspicious sign-in, what changed, what was sent, what was read. You will need it for the next step.

Hour 12 to 24: decide who needs to know

Three groups usually need a decision before the day is out.

  1. Your insurer. Check the policy’s notice requirements, and whether it names responders you must use, before you hire outside help, so the cost stays covered.
  2. Customers and suppliers in the thread. Anyone who received a fraudulent payment request should hear from you by phone, at a number you already have, that the request was not genuine.
  3. Counsel, about notification. If the mailbox held personal information, such as Social Security numbers, account numbers, or a username or email address with its password, a state breach-notification law may apply. New York’s SHIELD Act is one example. Whether a notice is required depends on what was in the mailbox and what the attacker could reach, which is why the timeline and the audit log matter.

Bring the account back carefully

When the investigation is done, Microsoft’s guidance is to reset the password again and then re-enable the account. If the account was used to send spam, Microsoft may have blocked it from sending, and an admin will need to remove it from the Restricted entities page. Before the user signs in again, make sure multifactor authentication is enforced and that only methods the user actually owns are registered.

What would have made this smaller

None of these is new, and each one shortens or prevents the day above:

  • Phishing-resistant MFA for admins and finance staff. Microsoft recommends it for admin accounts in the same guidance.
  • Block automatic forwarding to outside addresses by default, with exceptions approved one at a time.
  • Restrict which apps users can approve on their own, so a convincing consent prompt does not hand over the mailbox.
  • A payment-change rule that everyone follows: any change to bank details is confirmed by phone, at a number already on file, before money moves.
  • Longer log retention if you can justify it, and a quarterly look at sign-in logs for anything odd.

When to get help

If this is happening now, contain the account and call the bank first, then bring in help for the investigation. If it has not happened yet, an hour spent checking forwarding rules, consent settings and log retention will make the bad day shorter.

Want to talk it through? Book a free 30-minute call with Security Medic Consulting.

Sources