Measuring the collaboration and remote-support attack surface together
The Teams campaign described by Arctic Wolf crosses three boundaries: a collaboration service, a tenant identity, and an endpoint support tool. A security program that measures only one boundary will miss the relationship that makes the attack work [1].
Teams external access creates the contact path. A newly created or acquired Microsoft 365 tenant supplies an external identity. Mail bombing creates urgency. Quick Assist or a commercial RMM product turns the conversation into control of a device. The reported chain reached payload execution in under 15 minutes in some cases [1].
What can be measured externally
ZoomEye observations can describe public service and product footprints. On November 18, 2026, successful all-scope queries returned:
| Query | Observation | Scope |
|---|---|---|
domain="teams.microsoft.com" |
202 | Public Teams domain observations |
domain="onmicrosoft.com" |
221 | Public tenant-domain observations |
app="AnyDesk" |
466 | Public AnyDesk product observations |
app="ScreenConnect" |
102,705 | Public ScreenConnect product observations |
http.body="Quick Assist" |
361 | Public page-text observations |
These measurements do not prove external access, trial status, malicious use, compromise, or victimhood. They show that the products and domains have a public footprint that can be inventoried and compared over time. The exact queries are linked in the references [3]-[7].
What must be measured internally
The organization needs an approved-domain inventory, a list of external-access exceptions, the number of first-time external tenants contacting employees, unmanaged-account policy, trial-tenant policy, and the users affected by each policy. It also needs a complete remote-support inventory with owners, approved use cases, installation paths, and logging coverage.
Connect those data sets with a common identifier. A tenant domain can be compared with the approved-domain list. A Teams chat can be joined to mail activity and endpoint processes by user and time. A remote-management installation can be tied to an owner and a change record. This is more useful than counting alerts without knowing what decision follows.
Review metrics that lead to action
Track first-seen external tenants, IT-themed sender names, external chats to high-risk users, mailbox bursts that coincide with Teams contact, unapproved RMM installations, and support sessions without a ticket. Set review periods and name the owner for each exception.
Microsoft's external-access controls include domain allowlists, blocklists, unmanaged-account settings, and trial-tenant settings [2]. Arctic Wolf recommends restricting external domains, blocking unmanaged and trial accounts, limiting Quick Assist and unsanctioned RMM tools, and teaching users to check the External tag and sender domain [1]. These recommendations are useful because they change the opportunity for contact or control.
Close the loop
An external scan should produce an inventory question, not a verdict. A Teams domain observation can prompt a review of collaboration exposure. An RMM fingerprint can prompt a check of approved installations. A tenant-domain observation can prompt a check of partner and provisioning records. The internal system answers whether the observation matters.
That approach keeps evidence separate from inference. It also gives the security team a way to show progress: fewer unknown external domains, fewer unexplained remote-support sessions, and a known response path when a user reports the next IT-support impersonation.
References
[1] Arctic Wolf, "Beyond the Inbox: How Threat Actors Abuse Microsoft Teams External Chat for Initial Access." https://arcticwolf.com/resources/blog/threat-actors-abuse-microsoft-teams-external-chat-for-initial-access/
[2] Microsoft Learn, "Manage external meetings and chat with people and organizations using Microsoft identities." https://learn.microsoft.com/en-us/microsoftteams/trusted-organizations-external-meetings-chat
[3] ZoomEye observation for domain="teams.microsoft.com", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=ZG9tYWluPSJ0ZWFtcy5taWNyb3NvZnQuY29tIg%3D%3D
[4] ZoomEye observation for domain="onmicrosoft.com", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=ZG9tYWluPSJvbm1pY3Jvc29mdC5jb20i
[5] ZoomEye observation for app="AnyDesk", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=YXBwPSJBbnlEZXNrIg%3D%3D
[6] ZoomEye observation for app="ScreenConnect", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=YXBwPSJTY3JlZW5Db25uZWN0Ig%3D%3D
[7] ZoomEye observation for http.body="Quick Assist", collected November 18, 2026. https://www.zoomeye.ai/searchResult?q=aHR0cC5ib2R5PSJRdWljayBBc3Npc3Qi
Top comments (0)