Skip to content

Response, New York. Noah Park: Check Point Patches Actively Exploited Management Server Path Traversal. Detection, Dallas. Ava Okello: Report Finds Nearly Half of Deployed Detections Need Attention Before SOCs Can Trust Them.

SOCtember

Always First. Fast SOC News.

Diagram from Arctic Wolf titled Attack Chain showing nine numbered steps of a Microsoft Teams help desk scam: chat created from an external IT Support tenant, a vishing call, remote access through Quick Assist or ScreenConnect, a reconnaissance burst, a payload drop, a C2 beacon, a second RMM tool, lateral movement and finally exfiltration and ransomware.
Diagram: Arctic Wolf Labs, "Beyond the Inbox: How Threat Actors Abuse Microsoft Teams External Chat for Initial Access," Oct. 26, 2026 (fair use). The attack chain Arctic Wolf reconstructed from its cases.

Detection

Arctic Wolf Counts More Than 400 Incidents in Which Attackers Posed as IT Support in External Microsoft Teams Chats, and Says the ChatCreated Event Is the Earliest Place to Catch Them

The pattern rarely changes: an inbox flood, then a one-on-one Teams message from an outside tenant named like the help desk, then Quick Assist or a commercial remote tool. Arctic Wolf says the first chat, logged by Microsoft as ChatCreated, fires before the employee has done anything, and that attackers are moving from free onmicrosoft tenants to cheap .top domains.

Ava Okello, DetectionLondon6 min read

LONDON - The message arrives inside Microsoft Teams, not in email, and it comes from someone calling themselves IT support. Often it lands minutes after the employee's inbox has filled with hundreds of newsletter sign-ups. The sender offers to fix the problem. Within a quarter of an hour, if the employee plays along, someone else is running code on the laptop.

Arctic Wolf Labs said on Monday that it has analyzed more than 400 incidents in which attackers used Teams' external chat feature for impersonation scams or to deliver malicious files, and that the same habits keep showing up: the same throwaway tenants, the same "IT Support (Internal)" style display names, the same remote-access tools and, increasingly, the same payload domains. The report, by researcher Shubham Dubey, stresses that no Teams vulnerability is involved. External access, also called federation, is on by default, which lets any user in another Microsoft 365 tenant find, message and call an organization's employees when both tenants allow it.

The technique is not new. Microsoft documented the chain of email bombing, a fake help desk on Teams and Quick Assist in its reporting on Storm-1811, linked to Black Basta, in mid-2024, and researchers have since tied it to Cactus, 3AM, Chaos and Zscaler's GoGRPC campaign. What the new report adds is volume and a detection point.

How the fake help desk is built

Between 30 and 40 percent of the cases Arctic Wolf analyzed used a default *.onmicrosoft[.]com tenant, which needs no verification and can be set up in minutes. The tenant names lean on support, IT, admin, network or help desk words, sometimes with 365, online, portal or a number added, such as helpdeskteam365, itcoreportal, supportedge33 and networkadmin916.

The display name does the persuading. Attackers append "(Internal)" so the sender reads as in-house even while Teams labels them External. They add friendly first names to look like a real service desk employee, localize the wording for French or Spanish speakers, and used more than 100 shield, tool or check-mark emoji markers in display names, which Arctic Wolf says may point to AI-assisted automation. Most recently, some put the target company's own name or domain in the display name.

Screenshot of Arctic Wolf's appendix table listing display names and patterns used by the attackers, including Help Desk, IT Support, HelpDesk Manager, IT Service Desk with first names such as Liam, Matt and Scott, HelpDesk Supervisor, Enterprise IT and Security (Internal), IT Tech Support Internal, Business IT Solutions and Support (Internal), Maintenance Service, IT-Representative and IT/Help Desk.
Screenshot: Arctic Wolf Labs, "Beyond the Inbox," appendix, Oct. 26, 2026 (fair use).Display names observed across the incidents.

The infrastructure is shifting. Among the paid domains Arctic Wolf identified over the past few months, about half used the .top top-level domain, about 40 percent used .com and just over 5 percent used multi-level domains such as .com.pl, .com.vn and .com.br. Some of those domains were set to redirect web visitors to Microsoft or to the victim's own site, so a curious analyst would see something legitimate.

From a chat to hands on the keyboard

The chat is only the door. Arctic Wolf assesses that the attacker's immediate goal in most cases is interactive access through a legitimate remote monitoring and management tool, often secured with a Teams voice call. Quick Assist was the most common, likely because it is built into Windows, signed by Microsoft and needs only a six-digit code read back by the victim. AnyDesk, ScreenConnect, TeamViewer, Atera, Zoho Assist, NetSupport and LogMeIn also appeared, and attackers who wanted to stay often installed a second commercial tool through msiexec so they would keep a channel after the first was cleaned up.

The pace is fast. Attackers typically hit several users within about five minutes, most likely through automation, and customers reported payload execution in under 15 minutes once a victim engaged. In several cases Kerberoasting and GoldenPAC-related credential-access detections fired alongside the first Teams alert, meaning the endpoint was already compromised by the time the chat was triaged. Attackers also tend not to paste links into Teams, where Safe Links scanning would see them; they browse to the payload themselves during the remote session.

In one case Arctic Wolf describes as potentially novel, the fake help desk had the user run a PowerShell command that pulled an MSI from Azure Blob Storage. About 30 seconds after launch, a script named cleaner.ps1 dropped into AppData\Local\PowershellCleaner\scripts and decoded an AES-encrypted loader, which unpacked a pythonw interpreter and a Python implant, ss.py, from archive6.zip. The implant queried Active Directory over LDAP for every user and computer object and sent the results to command servers picked at random from a list.

Screenshot of the Microsoft Teams admin center External access page showing organization settings and toggles for whether people can communicate with external Teams users, unmanaged Teams accounts and Skype users.
Screenshot: Arctic Wolf Labs, "Beyond the Inbox," Figure 1, Oct. 26, 2026 (fair use).The Teams External access settings attackers rely on.

Where to catch it

Arctic Wolf says its alerts most often fire at the first contact: the one-on-one chat from a new external tenant, which Microsoft records as a ChatCreated event, before the user has replied. It flags tenants that open chats with many customers at once, watches the top-level domains the campaigns favor, alerts on remote tools that are not sanctioned in a given environment and raises priority when an RMM install is followed by reconnaissance commands.

Screenshot of Microsoft's Teams and Skype for Business users in external organizations setting, with the option Allow only specific external domains selected.
Screenshot: Microsoft, via Arctic Wolf Labs, "Beyond the Inbox," Figure 3, Oct. 26, 2026 (fair use).The allow-list setting Arctic Wolf recommends.

What SOC teams can do now

  • Alert on Teams ChatCreated events from external tenants, especially new *.onmicrosoft[.]com tenants and display names containing IT, help desk, support or "(Internal)", and correlate them with an inbox flood in the previous hour.
  • Join that signal to endpoint telemetry: QuickAssist.exe or msra.exe launching, any RMM installer run through msiexec, PowerShell with an explorer.exe parent pulling from Azure Blob Storage, pythonw.exe running from AppData, and LDAP enumeration through adsisearcher.
  • Move Teams external access from allow all to an allow-list of partner domains, block chats from unmanaged Teams accounts, and block trial tenants with Set-CsTenantFederationConfiguration -ExternalAccessWithTrialTenants "Blocked".
  • Let the security team manage blocked domains so .top indicators can be pushed through the Defender Tenant Allow/Block List, and restrict Quick Assist and unsanctioned RMM tools to authorized IT roles.
  • Hunt the published infrastructure, including getitconnect[.]top, get-help-systems[.]top, helpdesk-sys[.]top, update-software[.]top, sec-check-it[.]top and helpdesk-support-us3[.]com, and the tenants helpdeskteam365[.]onmicrosoft[.]com, itcoreportal[.]onmicrosoft[.]com and supportedge33[.]onmicrosoft[.]com. The full lists are in the report's appendix.

Sources:


Ava Okello covers detection engineering, EDR telemetry, and SOC hunting for SOCtember from London.

Related stories

Detection desk