June 5, 2026

    Microsoft Teams Helpdesk Impersonation Attack Explained

    THREAT ADVISORY

    Microsoft Teams Cross-Tenant Helpdesk Impersonation Leading to Credential Theft & Data Exfiltration

    SEVERITY

    High / Active Campaign

    PUBLISHED

    June 5, 2026

    CATEGORY

    Social Engineering / Initial Access

    PLATFORMS

    Microsoft 365 · Windows

    EXECUTIVE SUMMARY

    Blumira’s Security Operations team has been tracking an active campaign in which threat actors impersonate IT helpdesk personnel via Microsoft Teams to socially engineer victims into granting remote access through Quick Assist. We successfully replicated the full attack chain in a controlled lab environment to understand its behavior end-to-end, and have since detected and disrupted multiple real-world attempts before they could progress beyond initial access. The attack chain relies entirely on legitimate collaboration tools and native Windows protocols. This advisory documents how the technique works, what we observed, and what organizations should do now to reduce their exposure.

    01 Threat Actor — Attribution & Infrastructure

    In our lab replication of this attack, threat actors are operating Microsoft Teams accounts presenting as “Help Desk” from an external trial tenant, contacting multiple users over a 13-minute window. Trial tenants are deliberately chosen — they enable rapid account creation with no vetting. The indicators below are drawn from our lab environment and corroborated against real-world attempts we detected and stopped before they could progress through the full kill chain.

    Field

    Value

    Attacker Account

    ithelpdesk8@kilan.onmicrosoft[.]com

    Display Name

    Help Desk

    Source IP

    194.26.229[.]22 — Russia (VirusTotal: Flagged)

    C2 IP #1

    2600:1407:7400:18b::356e (port 80, cloud-hosted CDN range)

    C2 IP #2

    2606:4700:4408::ac40:9517 (port 80, Cloudflare range)

    Exfiltration

    Rclone → External Cloud Storage

    Note: The C2 IPv6 addresses had no VirusTotal hits at time of analysis. They appear to be cloud-hosted ranges used to blend C2 traffic into legitimate CDN activity. The initiating IP 194.26.229[.]22 is confirmed malicious.

    02 Intrusion Timeline

    All times UTC. The following timeline reflects our controlled lab replication of this attack, which allowed us to document the full kill chain end-to-end. In real-world detections, Blumira disrupted attempts at the initial access and reconnaissance stages — before credential dumping or lateral movement could occur. From first Teams contact to interactive endpoint access: 21 minutes.

    16:53 – 17:06

    Initial Teams Contact & Vishing

    Attacker account ithelpdesk8@kilan.onmicrosoft.com initiates Teams chats and voice/video calls with multiple users posing as IT helpdesk. In two calls the attacker transmitted live video. Recipients received unsolicited spam floods prior to contact — a deliberate lure to manufacture urgency and make the helpdesk call feel credible.

     

    17:14

    Quick Assist Access Granted

    Endpoint activity begins. A victim grants the attacker remote access via Quick Assist, giving full interactive desktop control. This step completes in under one minute after the user is walked through the approval prompts by the attacker.

     

    17:14 – 17:16

    Rapid Domain Reconnaissance

    net.exe group /dom is executed five times in under two minutes via PowerShell, enumerating domain groups. Windows Terminal (wt.exe) spawns PowerShell; Remote Desktop (mstsc.exe) is launched directly from PowerShell. This burst pattern is the clearest behavioral signal in the chain.

     

    Shortly After

    Credential Dumping & C2 Established

    PowerShell initiates outbound connections on port 80 to two IPv6 C2 addresses. Credential dumping occurs. Payload staging begins in C:\ProgramData with DLL sideloading via trusted binaries. Encrypted configuration is written to the user registry for persistence.

     

    Post-Pivot

    Lateral Movement, RMM Deployment & Exfiltration

    Using harvested credentials, the attacker pivots via WinRM (TCP 5985) to domain controllers and identity infrastructure. Commercial RMM software is installed remotely via msiexec.exe for persistent secondary access. Rclone exfiltrates targeted documents to external cloud storage with file-type exclusions to minimize detection.

    03 Intrusion Playbook — Stage by Stage

    This campaign follows a structured, human-operated playbook documented independently by Microsoft’s Defender Security Research Team across multiple intrusions. Each stage builds on the last using legitimate tools and native protocols.

    #

    Stage

    Description

    1

    Initial Access

    Attacker contacts victims from an external M365 trial tenant displaying as “Help Desk.” Spam floods arrive first to create urgency, then a Teams message or call follows. Microsoft Teams external-sender warnings appear at each step — the attack succeeds only when users bypass them.

    2

    Remote Access via Quick Assist

    The victim is convinced to open Quick Assist, enter an access code, and approve elevation prompts. Full interactive desktop access is established in under one minute.

    3

    Rapid Reconnaissance

    Within 30–120 seconds: domain group queries (net group /dom), privilege checks (whoami /all), network discovery (ipconfig /all, arp -a), and OS build queries via registry.

    4

    DLL Sideloading

    A staging bundle drops into C:\ProgramData. Legitimate signed executables — AcroServicesUpdater2_x64.exe, ADNotificationManager.exe, DlpUserAgent.exe — sideload attacker-supplied DLLs, running malicious code under trusted signatures.

    5

    Command & Control

    Sideloaded component beacons outbound HTTPS to attacker infrastructure. Encrypted C2 config stored in user-context registry (not disk) to reduce forensic visibility.

    6

    Lateral Movement

    Using harvested credentials, attacker pivots via WinRM (TCP 5985) to domain controllers and identity infrastructure — credential-backed, protocol-native, blends into normal admin traffic.

    7

    Exfiltration

    Commercial RMM software installed via msiexec.exe for persistent secondary access. Rclone systematically transfers targeted documents to external cloud storage with file-type exclusions.

     

    04 Indicators of Compromise

     

    The following indicators were captured during our lab replication of this attack and corroborated against real-world detection telemetry. IPs are defanged with brackets to prevent accidental clicks. Network defenders should hunt for these patterns within their environments.

    Type

    Indicator

    Context

    IP — Attacker

    194.26.229[.]22

    Source of all Teams messages; Russia-geolocated; VirusTotal flagged

    IP — C2

    2600:1407:7400:18b::356e

    Outbound PowerShell C2, port 80; cloud-hosted CDN range

    IP — C2

    2606:4700:4408::ac40:9517

    Outbound PowerShell C2, port 80; Cloudflare range

    Account

    ithelpdesk8@kilan.onmicrosoft[.]com

    Attacker Teams account; display name: “Help Desk”; cross-tenant trial

    Process

    QuickAssist.exe → cmd.exe / powershell.exe

    Quick Assist spawning shell — high-fidelity malicious indicator

    Command

    net.exe group /dom (×5 in <2 min)

    Domain group enumeration burst immediately post-access

    Process

    wt.exe → powershell.exe

    Windows Terminal spawning PowerShell in attacker session

    Command

    mstsc.exe (from PowerShell)

    RDP launched from attacker-controlled shell

    Process

    AcroServicesUpdater2_x64.exe

    Sideload host; beacons to C2 over HTTPS; staged in ProgramData

    Process

    rclone.exe

    Targeted data exfiltration to external cloud; treat any execution as critical

     

    05 Recommended Actions

    Blumira is actively alerting on these reported actions, but for security practitioners outside of the Blumira customer-base, please see our steps below:

    Priority Notice: Organizations using Microsoft Teams with default external access settings are currently exposed. User education and Teams external access restrictions are the two highest-leverage changes and should be implemented immediately.

    01. Restrict Microsoft Teams external access

    By default, Teams allows communication with all external Microsoft 365 tenants. In the Teams Admin Center (Users → External Access), scope this to an explicit allowlist of trusted partners, or block all external domains if cross-tenant collaboration is not required.


    Reference: learn.microsoft.com/en-us/microsoftteams/trusted-organizations-external-meetings-chat

    02. Block federation with trial-only tenants

    Run: Set-CsTenantFederationConfiguration -ExternalAccessWithTrialTenants "Blocked". In both our lab replication and real-world detections, attacker accounts operated from trial tenants. This PowerShell setting is not enabled by default and closes a common initial access vector.

    03. Harden Microsoft Entra ID External Identities settings

    Restrict guest invite permissions to the Guest Inviter role, limit external collaboration to approved domains, and scope SharePoint/OneDrive sharing to existing guests only. Reference: learn.microsoft.com/en-us/entra/architecture/9-secure-access-teams-sharepoint

    04. Educate users — the single most important control

    Train employees to: treat unsolicited Teams contact claiming to be IT helpdesk as suspicious; always check the external sender label; understand that real IT will never contact them from an @*.onmicrosoft.com domain; know that spam floods followed by a Teams call is a compound attack indicator. Consider establishing a verbal helpdesk passphrase.

    05. Restrict or monitor Quick Assist and RMM tools

    Quick Assist should only be used when IT explicitly initiates it. If unused, block the binary via policy. Alert immediately on QuickAssist.exe spawning child processes (cmd.exe, powershell.exe) — this is a high-confidence malicious indicator.

    06. Enable Safe Links for Teams and Zero-Hour Auto Purge (ZAP)

    Microsoft Defender for Office 365 can inspect links in Teams conversations at time-of-click and retroactively quarantine malicious messages. Confirm both features are active in your tenant.

    07. Enforce MFA and Conditional Access for administrative roles

    Lateral movement in this campaign was credential-backed. Phishing-resistant MFA (FIDO2/passkeys) combined with Conditional Access requiring compliant devices for WinRM and domain controller access significantly raises the cost of post-compromise pivoting.

    08. Alert on Rclone and unauthorized sync tooling

    Rclone is not a common enterprise tool. Any execution of rclone.exe — especially with --config rclone_uploader.conf or high-concurrency transfer flags — should trigger an immediate critical alert. Hunt for it proactively in your environment now.

     

    06 References

    Microsoft Security Blog — Cross-tenant helpdesk impersonation to data exfiltration: A human-operated intrusion playbook (April 18, 2026)

    https://www.microsoft.com/en-us/security/blog/2026/04/18/crosstenant-helpdesk-impersonation-data-exfiltration-human-operated-intrusion-playbook/

    Microsoft Entra — Secure external access to Microsoft Teams, SharePoint, and OneDrive

    https://learn.microsoft.com/en-us/entra/architecture/9-secure-access-teams-sharepoint

    Microsoft Teams Admin — Manage external meetings and chat with trusted organizations

    https://learn.microsoft.com/en-us/microsoftteams/trusted-organizations-external-meetings-chat?tabs=organization-settings

    © 2026 Blumira, Inc. — blumira.com

     

    Nick Dixon

    Nick is a cybersecurity professional with over a decade of experience in IT security and operations management. A Detroit native and graduate of Eastern Michigan University's Information Assurance program, he currently serves as Security Analysts & Technical Support Manager at Blumira, where he has advanced through...

    More from the blog

    View All Posts