TLDR: Microsoft 365 had a long outage on November 12 and 13. Microsoft says it was not a cyberattack. Attackers used it anyway: on November 13, Microsoft Threat Intelligence detected phishing emails dressed as Microsoft 365 service health notices that cited the real Exchange Online incident ID, EX1550543, and warned that mail held during the outage would be deleted within 24 hours unless the recipient clicked to restore it. The same quarter, Microsoft saw malicious Teams callers drop explicit “help desk” names in favor of ordinary personal names, and Canada’s Cyber Centre warned that fake IT callers may hang up when an employee tries to verify independently. The common thread is that a lure can now get every detail right. For MSSPs, the useful response is to teach verification by channel, not by content: pre-write the outage message before the outage, give users one place to check that they open themselves, make a callback to the number on file the rule for any request involving credentials or remote access, tighten the Teams and remote-tool settings that make the call possible, and count how many people report the fake.
Last Thursday, a little after 2 PM in Bogotá, a teammate posted in our team chat: “Howdy team, not sure if any of you are affected by this, but M365 seems to be having some issues at the moment.” Another teammate answered that Outlook was not working and that they were following it on Hacker News. The thread there, titled “Service degradation on Microsoft 365,” went past 1,100 points.[1]
It turned into a long one. Microsoft’s own record puts the impact from 5:05 AM Eastern on Thursday to 11:10 AM Eastern on Friday, and Downdetector peaked at nearly 130,000 Outlook reports just after 1 PM Eastern on Thursday.[2] Exchange Online and Outlook took the worst of it, with possible effects on Teams, SharePoint Online and Microsoft 365 Copilot. Microsoft said plainly that it was “a technical issue not related to cyber or AI / agents."[3]
The attackers did not need it to be a cyberattack. They only needed it to be real.
The notice was accurate
On November 13, Microsoft Threat Intelligence detected a phishing campaign that impersonated Microsoft 365 service health notices.[3] The emails referred to EX1550543, the actual incident identifier for the Exchange Online outage.[3] Some told recipients that email held during the outage would be deleted within 24 hours unless they clicked a link to restore it.[3]
The details were tailored. Subject lines started with “Outage -” followed by a six-character code. Sender display names combined organization-specific wording with “Outage Notification” or “Service Health."[3][4] The emails came from three unrelated domains and used recipient-specific SendGrid click-tracking links that redirected to a third-party site.[3][4] Microsoft described the activity as consistent with credential phishing, said credential collection and malware delivery had not been confirmed, and reported that its security products were blocking it.[3]
Its advice for security teams was to examine suspicious emails and links, investigate user clicks, review Safe Links policies, and verify service incidents directly through the Microsoft 365 admin center.[3]
One person on Hacker News reported that the status page would not even load.[1] The fake notice, meanwhile, had the right incident number. For part of that day, the most precise outage notice some people got was the phish.
Accurate details are cheap
Security awareness has spent years teaching people to look for mistakes. A misspelled brand. A generic greeting. A logo that is slightly off. A story that does not match anything happening in the user’s world.
This campaign had none of those problems. The outage was real. Everyone in the building already knew Outlook was broken. The incident ID was public within hours, because Microsoft, admins and journalists were all repeating it. An email that said “your mail was held during the outage” was describing something that had just happened to the reader. The only invented part was the deadline, and the deadline is what made people move.
iTnews made the point well: tailoring like this makes the messages harder to ignore, “particularly in inboxes already full of genuine outage updates."[4]
The same shift is visible on the phone and in Teams. Microsoft’s report on the third quarter of 2026 counted about 141,000 malicious Teams call attempts.[5] Attackers moved away from display names that announced an IT role: ordinary personal names rose from 46% of the observed July sample to 88% in September, while explicit help desk identities fell from 17% to 2%.[5] Microsoft notes that a personal name can still come with a help desk story, and that calls clustered in recipients’ weekday working hours, when someone was likely to pick up.[5] It also cautions that these are attempts, many of them unanswered, rejected or short, not successful attacks.[5]
Canada’s Canadian Centre for Cyber Security described the live version in an October alert about an ongoing help desk vishing campaign.[6] Callers impersonate internal IT and claim an urgent passkey migration, MFA enrollment or password change is required. They may spoof legitimate help desk numbers, reference internal IT staff or missed deadlines, and then send the employee to an organization-branded sign-in page that relays credentials and MFA responses to the real identity provider.[6]
And one line in that alert deserves to be on every training slide this quarter. The callers “may end the call if the employee attempts independent verification."[6]
Verify the channel, not the content
Put the outage phish and the help desk call next to each other and the lesson is the same. Checking whether the details are right no longer protects anyone, because the details can all be right. The incident number is real. The outage is real. The IT colleague the caller mentions may be real too.
What an attacker cannot control is a channel the user starts. A status page typed into the address bar. The admin center opened from a bookmark. A callback to the help desk number printed on the company intranet. When the user opens the channel, the attacker is not on the other end, which is exactly why the CCCS callers hang up when someone tries.[6]
That is the habit we keep coming back to at Symbol, and it is boring on purpose: verify through a channel you started. It works on an email with a perfect incident number, on a Teams call from an ordinary name, and on a text that says the help desk needs a code. It does not require anyone to notice a mistake.
What MSSPs can set up before the next outage
The next big outage will happen. When it does, your clients will be confused, impatient and more willing than usual to click a button that promises to fix things. Most of the useful work happens before that day.
1. Pre-write the outage message
Write the two-line note now and get it approved: “During any Microsoft outage, we will never email you a link to restore mail or re-enter your password. Status updates come from [your channel], and you can always call us at [number on file].” Send it to clients as a standing policy, and send it again within the first hour of the next major incident. In this case, Microsoft detected the fake notices the day after the outage started.[3] An early message reaches people before the lure does.
2. Give users one place to check, and make them open it
Microsoft’s own guidance is to confirm service incidents in the Microsoft 365 admin center, not in an unexpected email.[3] Most employees will never open the admin center, so give them the equivalent: your portal, your status page or your support line. The important part is that they type it or open it from a bookmark. A link inside a notice, even a correct one, trains the wrong reflex.
3. Make the callback the rule, and make hanging up acceptable
The CCCS recommends instructing employees to independently verify unexpected IT calls that ask for MFA, passkey or password changes, regardless of caller ID or urgency, and to strengthen verification for MFA resets, account recovery and new device enrollment.[6] Write it as one sentence your clients’ staff can remember: if someone contacts you first and asks for a code, a password, an MFA change or remote access, hang up and call back the number on file.
Your own help desk needs the mirror image. “I have a new phone, send the code here instead” defeats verification because it verifies through the channel the caller supplied.[7] Verify against a device or address registered before the request, and escalate to a person when that is not possible.[7]
4. Shrink the surface the call depends on
Behavior is the product, but settings make the behavior easier. eSentire’s advisory on the current Teams vishing activity recommends disabling communication with unmanaged Teams accounts and allowing external Teams messages and calls only with named partner domains, standardizing on one approved remote-assistance tool and blocking the rest, and blocking device code flow through Conditional Access.[8] The CCCS adds phishing-resistant MFA, particularly for privileged and high-risk users.[6]
Microsoft is also rolling out AI phishing detection for first-time chats with external contacts in Teams, designed to catch impersonation that contains no link or attachment. Worldwide availability is expected by mid-December 2026, and no admin setup is required.[9] It is a welcome layer. It is still a warning on a screen, and the habit has to be there when the warning is not.
5. Hunt the click, then count the reports
When a lure like this lands, Microsoft points security teams to Defender for Office 365 Threat Explorer and to advanced hunting across the EmailEvents, EmailUrlInfo and UrlClickEvents tables to find messages, examine URLs and trace clicks.[4] A click is a lead to investigate, not proof that a password left the building.
Then look at the other number. How many people reported the fake outage notice, and how quickly after it arrived? A user who reports in the first ten minutes protects everyone who has not opened it yet. Report rate tells you whether the program is working. Click rate only tells you who had a bad afternoon.
6. Rehearse the shape after the dust settles
Do not run a simulation in the middle of a real outage, when people need genuine notices to get their work done. Afterward, the shape is worth rehearsing: a service notice with a deadline and a restore button, or a message from an ordinary-sounding name asking to “check something quickly” on a call. The lesson you want people to practice is not “spot the typo.” It is “open the channel yourself.”
What not to over-read
Microsoft has not confirmed that this campaign collected credentials, and it says its products blocked it.[3] The Teams numbers are call attempts, not successful compromises.[5] Nobody should read this as proof that outages cause breaches.
What it does show is speed and accuracy. Within a day of a global outage, someone had a polished, tailored lure that referenced the correct incident. That will happen again with the next outage, the next breach notification and the next vendor migration, because accurate details are now the easy part.
Where Symbol Fits
Symbol exists to help admins run security awareness across many tenants without staffing a person for every client, and this is the kind of week the program is for.
The phishing simulation library is built from active campaigns and takes custom template requests, so you can rehearse the service notice shape and the first time a client’s staff sees a deadline and a restore button is not the real one. The security awareness training you assign across tenants is where the one-sentence rule lands: verify through a channel you started. And Report a Phish gives every user a single action for the notice that looks right but arrived from the wrong place, and puts the report in front of your analysts instead of in someone’s deleted items.
The phish had the real incident number. Make sure your clients’ staff know the number that matters is the one already on file.
References
- Hacker News, “Service degradation on Microsoft 365,” November 12, 2026. https://news.ycombinator.com/item?id=50471731
- Rich Ord, “Microsoft 365 Outage Knocks Out Outlook for More Than a Day; Microsoft Apologizes and Cites a Network Upgrade Issue,” WebProNews, November 12, 2026, updated November 16, 2026. https://www.webpronews.com/microsoft-365-outlook-outage-servers-health-checks-no-root-cause/
- SecurityWeek, “Microsoft Rules Out Cyberattack in Outage, Warns of Phishing Campaign,” November 16, 2026. https://www.securityweek.com/microsoft-rules-out-cyberattack-in-outage-warns-of-phishing-campaign/
- iTnews, “Phishers quick to exploit extended M365 outage,” November 2026. https://www.itnews.com.au/news/phishers-quick-to-exploit-extended-m365-outage-630347
- Microsoft Threat Intelligence, “Email threat landscape: Q3 2026 trends and insights,” Microsoft Security Blog, November 5, 2026. https://www.microsoft.com/en-us/security/blog/2026/11/05/email-threat-landscape-q3-2026-trends-and-insights/
- Canadian Centre for Cyber Security, “AL26-028: Ongoing Help Desk Voice Phishing Campaign Targeting Canadian Organizations,” October 28, 2026. https://www.cyber.gc.ca/en/alerts-advisories/al26-028-ongoing-help-desk-voice-phishing-campaign-targeting-canadian-organizations
- Mathieu Tougas, “MSP Help Desk Social Engineering: Five Requests to Treat as Suspicious,” Mizo, November 5, 2026. https://mizo.tech/blog/msp-help-desk-social-engineering-ai-verification/
- eSentire Threat Response Unit, “Increase in Impersonation Vishing Attacks Leading to Data Extortion,” security advisory, November 3, 2026. https://www.esentire.com/security-advisories/increase-in-impersonation-vishing-attacks-leading-to-data-extortion
- TechRepublic Staff, “Microsoft Teams AI Phishing Detection: Rollout and Deepfake Warnings,” TechRepublic, November 11, 2026. https://www.techrepublic.com/article/news-microsoft-teams-ai-phishing-detection-deepfake-warnings/
