Skip to navigation

Threat reportPhishingTL-2026-2276

"Spring Ring" Vishing Campaign Abuses Microsoft Teams, Quick Assist, and PetitPotam for NTLM Relay

highACTIVE

Unit 42 (Palo Alto Networks) documented a vishing campaign dubbed 'Spring Ring' running January-April 2026, in which 26+ attacker-controlled Microsoft Teams identities posed as IT support to socially engineer 150+ employees across 10+ companies into granting remote access via Quick Assist/RMM tools or executing malicious files, leading to AMSI/Defender bypass and PetitPotam-based NTLM relay attempts against domain controllers.

Key facts

IdentifiersTL-2026-2276, CVE-2021-36942
Also known asSpring Ring
SeverityHIGH
First disclosed
Exploitation statusLinked CVE exploited in the wild; Listed in CISA KEV: CVE-2021-36942 (added )
EPSSCVE-2021-36942: 66.0% (99.2th percentile)
Affected productsMicrosoft Microsoft Teams; Microsoft Windows Server Active Directory Domain Controllers; Microsoft Quick Assist
ToolsPetitPotam, Quick Assist
MITRE ATT&CK techniquesT1036 Masquerading, T1046 Network Service Discovery, T1053 Scheduled Task/Job, T1059.001 PowerShell, T1069.002 Domain Groups, T1071.001 Web Protocols, T1176 Software Extensions, T1187 Forced Authentication, T1204.002 Malicious File, T1219 Remote Access Tools, T1550 Use Alternate Authentication Material, T1557.001 Name Resolution Poisoning and SMB Relay and 5 more in the attack flow
MotivationUNKNOWN
Target sectorsmultiple-industries
Detection rules9 (1 shown free)
Sources7 cited sources

Detection coverage

As of 2026-09-01, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2276 across Splunk SPL, Microsoft KQL and Sigma, covering 33 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.

Free detection: External Teams Vishing-to-Quick-Assist Remote-Control Pivot

Detects the Spring Ring initial-access pattern: an external, unverified Microsoft Teams tenant with an IT/helpdesk-themed onmicrosoft.com identity (e.g. helpcenter@ithelpcenter365.onmicrosoft.com) initiates a chat/call with an internal user, followed within 15 minutes by that same user launching Quick Assist (QuickAssist.exe) or a common third-party RMM tool — the chat-to-call-to-remote-access transition Unit 42 flagged as the key behavioral signal across 150+ targeted employees.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1566.004, T1204.002, T1219, T1583.006, T1585
  • Data sources: o365_management_activity, sysmon, windows_process_creation
  • Severity: high
  • Confidence: high

Splunk SPL

index=o365 sourcetype="o365:management:activity" Workload=MicrosoftTeams Operation IN ("ChatCreated","MessageSent")
| rex field=UserId "@(?<sender_domain>[a-zA-Z0-9.-]+\.onmicrosoft\.com)$"
| where isnotnull(sender_domain) AND match(sender_domain, "(?i)(help|it|admin|support|desk|monitoring)")
| eval teams_time=_time
| join type=inner target_user
    [ search index=sysmon sourcetype="XmlWinEventLog:Microsoft-Windows-Sysmon/Operational" EventCode=1
      (Image="*\\quickassist.exe" OR Image="*\\AnyDesk.exe" OR Image="*\\TeamViewer*.exe" OR Image="*\\ScreenConnect*.exe")
    | rename User as target_user, _time as rmm_time, Image as rmm_image, ComputerName as host ]
| eval delta_sec=abs(rmm_time-teams_time)
| where delta_sec<=900
| table teams_time, sender_domain, target_user, host, rmm_image, rmm_time, delta_sec
| sort - teams_time

Microsoft KQL

let externalTeamsChats = CloudAppEvents
| where Timestamp > ago(24h)
| where Application == "Microsoft Teams"
| where ActionType in ("ChatCreated", "MessageSent")
| extend SenderDomain = tostring(split(AccountObjectId, "@")[1])
| where SenderDomain endswith "onmicrosoft.com" and SenderDomain matches regex @"(?i)(help|it|admin|support|desk|monitoring)"
| project TeamsTime = Timestamp, TargetUser = AccountDisplayName, SenderDomain;
DeviceProcessEvents
| where Timestamp > ago(24h)
| where FileName in~ ("quickassist.exe","AnyDesk.exe","TeamViewer.exe","ScreenConnect.exe")
| project RmmTime = Timestamp, TargetUser = AccountName, FileName, DeviceName
| join kind=inner externalTeamsChats on TargetUser
| where abs(datetime_diff('second', RmmTime, TeamsTime)) <= 900
| project TeamsTime, RmmTime, TargetUser, SenderDomain, FileName, DeviceName
| sort by TeamsTime desc

Sigma

title: Quick Assist Remote-Control Session Initiated (Possible Vishing Pivot)
id: 7f3b9c2e-1a44-4e9d-9c31-2b6a7d4f10a1
status: experimental
description: Detects execution of QuickAssist.exe or common third-party RMM tools; intended to be correlated within 15 minutes of an external Microsoft Teams chat/call from an unverified onmicrosoft.com tenant, as observed in the Spring Ring vishing campaign.
references:
    - https://unit42.paloaltonetworks.com/spring-ring-voice-phishing-campaigns/
author: Threadlinqs Team
date: 2026-09-01
tags:
    - attack.initial-access
    - attack.t1566.004
    - attack.execution
    - attack.t1204.002
    - attack.command-and-control
    - attack.t1219
    - attack.resource-development
    - attack.t1583.006
    - attack.t1585
    - attack.campaign.spring_ring
logsource:
    category: process_creation
    product: windows
detection:
    selection:
        Image|endswith:
            - '\quickassist.exe'
            - '\AnyDesk.exe'
            - '\TeamViewer.exe'
            - '\ScreenConnect.exe'
    filter_it_staff:
        User|contains:
            - 'IT-Support'
            - 'HelpDesk'
    condition: selection and not filter_it_staff
falsepositives:
    - Legitimate IT helpdesk-initiated remote support sessions
    - Approved third-party RMM usage by authorized administrators
level: medium

Known false positives:

  • Legitimate IT helpdesk-initiated remote support sessions conducted through verified internal Teams tenants
  • Approved third-party RMM usage by authorized administrators unrelated to an external chat

Tuning notes: Maintain an allowlist of verified internal/approved external Teams tenant domains to suppress known-good helpdesk vendors; tighten the join window from 15 to 5 minutes in high-volume environments to reduce coincidental matches.

8 more detection rules (Splunk SPL, Microsoft KQL, Sigma) · unlock with TL Creds. Sign up free to unlock the detection rules or query them with the Threadlinqs MCP server (Purple)
  • Company-Branded S3 Payload Download and Execution (Campaign B Delivery) · unlock with TL Creds
  • Headless Edge with Sideloaded Extension and Bundled Python SMB Recon Component · unlock with TL Creds
  • PetitPotam-Style Forced NTLM Authentication to Domain Controller · unlock with TL Creds
  • Fraudulent IT-Themed onmicrosoft.com Teams Tenant Contact · unlock with TL Creds
  • Post-Quick-Assist Domain Discovery Commands (whoami /groups, net group /dom) · unlock with TL Creds
  • End-to-End Spring Ring Chain: SMB Recon to PetitPotam Coercion within 30 Minutes · unlock with TL Creds
  • AMSI Bypass, Domain Discovery and Renamed Persistence Drop (Spring Ring Post-Exploitation) · unlock with TL Creds
  • [indicator] C2 Beacon and Internal SMB(445) Reconnaissance · unlock with TL Creds

Indicators of compromise

As of 2026-09-01, Threadlinqs Intelligence tracks 33 indicators of compromise for TL-2026-2276.

Sample indicators (3 of 33)

  • IP address: 193.32.248[.]251
  • Domain: san-sid[.]com
  • File hash: 24ab9fe5d5be62d3bf055a0ca4508e8bca2996b6d78649dce8145d8a27bc1c5b

Defanged, TLP:CLEAR. Full set of 33 indicators: unlock with TL Creds.

33 indicators of compromise · unlock with TL Creds. Compare plans

Technical analysis

Between January and April 2026, Unit 42 tracked a coordinated voice-phishing (vishing) operation it named 'Spring Ring,' published August 31, 2026 (with Help Net Security, Cyberpress, and other outlets carrying follow-on coverage September 1, 2026). Attackers registered external Microsoft Teams tenants (onmicrosoft.com) with names engineered to impersonate internal IT departments. Unit 42 identified at least 8 generic role-based malicious identities — helpcenter@ithelpcenter365.onmicrosoft.com, helpdesk@itprotectiondepartment.onmicrosoft.com, helpdesk@newsystemmaintenance.onmicrosoft.com, helpdesk@officedesk365.onmicrosoft.com, helpdesk@officesecures.onmicrosoft.com, itadmin@mandatorynetworkmonitoring.onmicrosoft.com, internal@internalusahelpdeskit.onmicrosoft.com, and ithelp@internalsystemsdaily.onmicrosoft.com — plus 17+ additional persona-based identities using personalized (rather than generic help-desk) display names, for a total of 26 distinct attacker identities. Operators initiated unsolicited Teams chats with employees and rapidly escalated to voice calls; unsuccessful attempts lasted seconds to a missed call, while successful engagements typically ran 10-15 minutes, and Unit 42 observed attackers cycling through 5-6 identities within minutes when a target didn't answer, targeting the chat-to-call transition speed as a key behavioral signal.

In the 'remote access' path (Campaign A), the caller talked the victim into launching Quick Assist (QuickAssist.exe) or a third-party RMM tool and granting control of the machine. Once in control, the attacker ran discovery commands (`whoami /groups`, `net group /dom`) to enumerate group membership and domain context, then delivered a heavily obfuscated PowerShell loader (retrieved via `Invoke-WebRequest`) — described by Unit 42 as using variable-manipulation and arithmetic obfuscation plus anti-analysis 'bloat' meant to time out deobfuscation tooling. The script disabled AMSI by forcing the `amsiInitFailed` flag and verified the bypass with a test scan, then encrypted host data and beaconed to the attacker domain `san-sid.com` with a 'tiny, nine-line' C2 stager to pull down further payloads (SHA256 `24ab9fe5d5be62d3bf055a0ca4508e8bca2996b6d78649dce8145d8a27bc1c5b`, fetched from `https://san-sid.com/owners`). Unit 42 reports this campaign was blocked by automated Cortex XDR Agent protections during the malware's execution phase.

In the 'malicious file' path (Campaign B), victims instead received a cloud-storage link — Unit 42 observed AWS S3 objects with URLs templated as `<company_name>-org-filters-update-<victim_name>.s3.us-west-2.amazonaws.com`, with the delivered executable named identically after the target company and the victim's own username. Running it moved the binary to `%TEMP%` and spawned renamed persistence copies (`vhlp-*.exe`, `scnr-*.exe`). The payload launched a hidden/headless Microsoft Edge instance with a sideloaded browser extension (browser hijacking) and a bundled Python component (observed at path `C:\ProgramData\IntegrityData\python.exe`) that performed internal SMB reconnaissance (TCP/445 port scanning) before invoking PetitPotam to coerce the domain controller into authenticating to an attacker-controlled listener — an NTLM-relay attempt aimed at escalating to domain-level privileges. Community technical write-ups estimate roughly 30 minutes elapsed from the initial Teams chat to the SMB scan/EFSRPC coercion attempt for a given victim. Unit 42 Managed Detection and Response blocked the domain-takeover attempt after PetitPotam's DC-coercion step; no report reviewed confirms a successful domain takeover in any observed instance. PetitPotam abuses the MS-EFSRPC `EfsRpcOpenFileRaw` API and is the technique behind the publicly tracked Windows LSA spoofing weakness CVE-2021-36942 (patched/mitigated by Microsoft in August 2021 via Extended Protection for Authentication / certificate-mapping guidance); Unit 42's write-up does not cite a CVE for the Spring Ring campaign itself since the underlying exposure is a systemic NTLM-relay/coercion design issue rather than a single unpatched flaw, and organizations that have not applied the 2021-era mitigations (EPA, SMB signing, disabling NTLM where feasible) remain relayable regardless of patch level.

Unit 42 published 16 attacker-side VPN/proxy IP addresses used for infrastructure, plus dozens of fraudulent onmicrosoft.com Teams-tenant identities. The report references 'Cloaked Ursa' (associated with APT29/Midnight Blizzard) as having previously run Teams-based credential-harvesting campaigns, explicitly as prior-art context for the collaboration-platform vector rather than as attribution — no specific threat-actor name, nation-state attribution, motivation, or CVE was assigned to Spring Ring itself. Help Net Security's coverage additionally notes that phishing alerts tied to collaboration platforms (Teams, Slack, etc.) rose to 42% of all phishing alerts in the first four months of 2026, up from roughly 30% in the preceding four-month period (approximately September-December 2025) — framing Spring Ring as part of a broader shift toward collaboration-platform vishing. Both Unit 42 and secondary analysis recommend correlating Teams audit logs, endpoint telemetry, authentication events, and network flow data within a single behavioral window (chat → call → RMM/PowerShell → cloud download → SMB scan → EFSRPC/NTLM) to reconstruct and detect the full attack chain, since no single log source captures it end to end.

MITRE ATT&CK techniques by tactic:

Defense Evasion

T1036 Masquerading

Discovery

T1046 Network Service Discovery; T1069.002 Domain Groups

Persistence

T1053 Scheduled Task/Job; T1176 Software Extensions

Execution

T1059.001 PowerShell; T1204.002 Malicious File

Command and Control

T1071.001 Web Protocols; T1219 Remote Access Tools

Credential Access

T1187 Forced Authentication; T1557.001 Name Resolution Poisoning and SMB Relay

Lateral Movement

T1550 Use Alternate Authentication Material

Initial Access

T1566.004 Spearphishing Voice

Exfiltration

T1567.002 Exfiltration to Cloud Storage

Resource Development

T1583.006 Web Services; T1585 Establish Accounts

defense-impairment

T1685 Disable or Modify Tools

Remediation

Immediate actions

  • Block the known Spring Ring attacker VPN/proxy IPs and the san-sid.com C2 domain at perimeter DNS/URL filtering and firewall egress controls
  • Restrict or disable inbound external Microsoft Teams chat requests from unverified onmicrosoft.com tenants, particularly those using IT/helpdesk-themed naming
  • Restrict Quick Assist and third-party RMM tool usage to approved IT staff and known support workflows; alert on RMM tool launches (e.g. QuickAssist.exe) by non-IT users
  • Hunt for the reported SHA256 (24ab9fe5d5be62d3bf055a0ca4508e8bca2996b6d78649dce8145d8a27bc1c5b), for spawned processes matching vhlp-*.exe / scnr-*.exe, and for python.exe running from C:\ProgramData\IntegrityData\ across the environment

Workarounds

  • Monitor and alert on unsolicited external Teams communications and unknown/unexpected cloud-storage (e.g. S3) download URLs
  • Monitor internal SMB scanning (port 445) and unexpected forced-authentication/NTLM-relay traffic toward domain controllers

Longer-term hardening

  • Deploy identity-based/behavioral analytics on Teams tenant metadata to flag rapid chat-to-voice-call transitions, cycling through multiple external identities in a short window, and suspicious external tenant naming patterns
  • Enforce PetitPotam/NTLM-relay mitigations on all domain controllers: Extended Protection for Authentication (EPA), SMB signing, disabling NTLM where feasible, and blocking unauthenticated MS-EFSRPC/MS-RPRN coercion calls
  • Deploy EDR behavioral detections for AMSI-bypass patterns (amsiInitFailed manipulation), hidden/headless browser instances with sideloaded extensions, and scheduled-task-based persistence of renamed binaries
  • Build a ~30-minute cross-source correlation rule linking external Teams chat -> voice call -> RMM/PowerShell launch -> cloud-storage (S3) download -> internal SMB(445) scan -> EFSRPC/NTLM authentication for the same user, per Unit 42 and community analysis of the Spring Ring attack-chain timing
  • Run recurring user-awareness training specific to collaboration-platform vishing (Teams/Slack impersonation of IT support), given the reported rise in collaboration-platform phishing alerts from ~30% to 42% of all phishing alerts

Affected and fixed versions

  • Microsoft — Microsoft Teams
    Vulnerable versions: Default external/federated tenant chat configuration
  • Microsoft — Windows Server Active Directory Domain Controllers
    Vulnerable versions: Domain controllers without Extended Protection for Authentication / SMB signing / NTLM-relay mitigations applied
  • Microsoft — Quick Assist
    Vulnerable versions: Any version used without organizational access restrictions (abused as a legitimate remote-access tool, not a software flaw)

Timeline

  • Rapid7 publishes technical analysis of the PetitPotam attack chain (MS-EFSRPC coercion), the NTLM-relay technique later reused in the Spring Ring campaign's Campaign B path.
  • Microsoft addresses CVE-2021-36942 (Windows LSA Spoofing), the underlying weakness abused by the PetitPotam coercion technique later used in the Spring Ring campaign's NTLM-relay path.
  • End of the prior four-month baseline period (roughly September-December 2025) in which collaboration-platform (Teams/Slack) phishing alerts represented approximately 30% of all phishing alerts, the baseline against which the Spring Ring-era Q1 2026 share (42%) is compared.
  • Spring Ring vishing campaign activity begins; Unit 42 observes attacker-controlled Microsoft Teams identities initiating unsolicited IT-support chats (approximate start of the January-April 2026 observation window).
  • Observed Spring Ring campaign activity concludes at the end of the January-April 2026 window tracked by Unit 42, having reached 150+ employees across 10+ companies via 26+ distinct attacker Teams identities.
  • Unit 42 (Palo Alto Networks) publishes 'Spring Ring: An Inside Look at Voice Phishing Campaigns in Microsoft Teams,' detailing the campaign's TTPs, the full list of attacker Teams identities and VPN/proxy IPs, and confirmation that Cortex XDR/MDR blocked both the PowerShell RAT execution and the PetitPotam domain-takeover attempt in observed cases.
  • Help Net Security, Cyberpress, and dev.to publish coverage summarizing Unit 42's Spring Ring findings, including the rise of collaboration-platform phishing alerts to 42% of all phishing alerts in the first four months of 2026.

Sources cited

Other reports on CVE-2021-36942

Previous and next in phishing

Drafted by the Threadlinqs research pipeline from 7 sources; published ; last updated —.

Threadlinqs Intelligence — Real-Time Threat Detection Platform

// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level

Latest Threats