- Microsoft Threat Intelligence saw a high-volume campaign on November 13, 2026 that imitated Microsoft 365 service-health notices and claimed held mail would be deleted within 24 hours.
- Microsoft assesses it was likely credential phishing, but its evidence did not confirm credential capture, malware or compromise.
- Teach staff to check outages in the Microsoft 365 admin center, and plan account resets for anyone who may have entered a password after clicking.
Why an outage notice makes a good disguise
Much phishing asks people to trust a stranger. This campaign imitated something familiar: the system that reports when email is broken. An outage notice is a routine part of working life.
That is the idea worth taking from the campaign Microsoft Threat Intelligence reported. The lure is built to exploit routine. It looks like IT, and it adds a deadline that invites a fast click. Microsoft does not say how many people clicked. The design is the story here, not its results.
What Microsoft observed
Microsoft says it observed a coordinated, high-volume campaign on November 13, 2026. It targeted organisations in multiple sectors worldwide. The emails posed as Microsoft 365 service-health notices.
One sample referred to an Exchange Online incident on November 12. It said four messages were being held. Recipients had one day to restore them or lose them. A "RESTORE MESSAGES" button was the call to action.
Microsoft says the lure played on fear of losing business email. As our own illustration, anyone who lives in their inbox could feel that pressure. A finance clerk waiting on an invoice is one example. Microsoft names no job roles.
How the attack works
Microsoft describes a short chain. A fake service-health email arrives. The recipient clicks the button. A tracking link records the click and passes the visitor along. The visitor lands on reallifeaccounting[.]com, in some samples at the path /themelist0.html. The emails came from sender domains Microsoft lists as moments-notice[.]com, planetmaison[.]com and newbooker[.]com.
Microsoft assesses the aim was likely to steal credentials. That usually means a page made to look like a Microsoft sign-in, which collects whatever is typed into it. This is the likely pattern, not a confirmed finding. Microsoft did not see the final page.
The design also targets defences. A blocklist works by matching something fixed, such as a subject line or a web address. Here the subjects ended in a random token. Each recipient also got a unique link. Microsoft says such choices complicate simple pattern matching. One entry on a list cannot catch every copy.
Microsoft also found a fake email thread in the message body, placed out of the reader's sight. It reads this as an attempt to slip past mail filters. The messages named Exchange Online, Outlook, Teams, SharePoint and Copilot to look authentic.
What is known and what is not
Microsoft did not see the page where victims would land. It saw no password fields, no submitted data and no later account misuse. For that reason it calls the credential-theft aim an assessment.
The samples were link-based. Microsoft saw no QR codes, no unusual attachments and no sign of malware. It does not name a group or malware family behind the campaign.
Microsoft says mail in environments protected by Defender for Office 365 is blocked or quarantined. It reports no signs of compromise. The source speaks only about protected environments. It says nothing about organisations without that product.
Why the admin center works as a check
Microsoft advises confirming outages in the Microsoft 365 admin center, not through links in unsolicited mail. In our view, this works because of where the check happens. An attacker can copy how an email looks. The admin center is a separate channel that the email does not control.
Microsoft adds a simple test. A real health notice would not send people to non-Microsoft domains to get held mail back.
The lesson here is that defences resting on each employee judging each message can be tested whenever attackers pick up a notable event, as Microsoft notes they historically do. One trusted place to check lowers that load.
Questions to put to your security team
First, can the team search mail for the three sender domains named above, the subject pattern and reallifeaccounting[.]com, and see who clicked? Microsoft publishes sample hunting queries for Defender XDR customers.
Second, what happens when someone clicked? For anyone who may have typed a password, Microsoft's list starts with a new password. It then ends active sessions and refresh tokens, and checks which multifactor devices are registered.
Microsoft also suggests checking the account's sign-in history and app permissions it has granted. Look for new mailbox rules that forward mail, and for delegate access.
Third, how strong is the sign-in itself? Microsoft asks for multifactor sign-in, favouring methods that resist phishing, such as passkeys or hardware security keys.
Fourth, are the basics switched on? Microsoft points to Safe Links, Safe Attachments and zero-hour auto purge. The last acts after delivery. When new threat data arrives, it can pull back mail that already landed.
One caution from Microsoft: do not broadly block shared email-delivery infrastructure because of this campaign alone. Combine sender, content, link and behaviour signals.
A real outage notice tells you what is wrong. A fake one tells you what to click. Teach people to notice the difference.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: Microsoft Security.





