Back to Intelligence

The Com Vishing Campaigns: Defending Healthcare and Finance Against Help Desk Social Engineering, Cloud Data Theft, and Extortion

SA
Security Arsenal Team
November 22, 2026
13 min read

Unit 42's latest research, Dialing for Data: Inside the Com's High-Profile Heists, documents what many of us in IR have been living through firsthand: affiliates of the loosely organized threat collective known as The Com — the same ecosystem that produced Scattered Spider/UNC3944-style intrusions — are running highly effective voice phishing (vishing) campaigns against healthcare and financial services organizations. The playbook is brutally simple and devastatingly effective: call the help desk, impersonate an employee, convince an agent to reset a password or MFA factor, walk straight into the victim's cloud environment, pillage SaaS data stores, and extort.

What makes this campaign wave different from commodity phishing is that there is no malware dependency and no vulnerability to patch. The initial access vector is a phone call and a permissive identity workflow. Once inside, the attackers operate almost entirely with legitimate cloud APIs, sanctioned admin tooling, and commercially available remote access software — which means your EDR sees very little, and your perimeter sees nothing. Healthcare and finance are being deliberately targeted because both sectors hold high-value regulated data (PHI, financial records), tolerate zero downtime, and historically pay.

If your organization's identity reset workflows can be defeated by a confident caller with a plausible story and a few OSINT-sourced personal details, you are exposed right now. This post breaks down the attack chain as observed, and gives you detection content and hardening steps you can deploy this week.

Technical Analysis

Who Is Being Targeted

Per Unit 42's analysis, Com affiliates are deliberately selecting:

  • Healthcare providers and payers — EHR-integrated environments, large help desks, high staff turnover, and heavy reliance on travel/contract clinicians whose identities agents may not know personally.
  • Financial services firms — wealth management, insurance, and regional banking, where a single compromised identity can expose client portfolios and wire-transfer workflows.

Targeting is OSINT-driven. Attackers harvest employee names, titles, and reporting structures from LinkedIn, data broker sites, and prior breach corpora, then select victims likely to have privileged SaaS access — IT staff, Salesforce administrators, finance team leads.

The Attack Chain

1. Reconnaissance and pretext building. Operators gather employee PII from people-search sites and social media, then collect enough detail (employee ID formats, manager names, badge terminology) to pass casual verification. They study the target's help desk scripts — often publicly inferable from job postings that name the ITSM platform (ServiceNow, Zendesk) and IdP (Okta, Microsoft Entra ID).

2. Vishing the help desk. The actor calls the service desk — frequently during shift changes, weekends, or after-hours when staffing is thin and scrutiny is low. Pretexts are consistent: "I'm traveling, I lost my phone, I need my MFA reset before my shift/my calls with clients." Some operators spoof the target's internal caller ID or call from numbers that appear local. The objective is one of:

  • Password reset to an actor-controlled email
  • MFA factor reset or enrollment of a new factor (the actor's phone/authenticator)
  • In some cases, SIM-swap-enabling social engineering against the mobile carrier in parallel, so SMS/voice factors route to the attacker

3. Initial access and persistence. With a fresh factor enrolled, the actor authenticates to the IdP from their own infrastructure — typically commercial VPNs or residential proxies geolocated near the victim to defeat naive geo-fencing. Persistence is established by registering additional MFA methods, adding recovery emails, and in some cases creating OAuth grants or app passwords that survive a later password reset.

4. Reconnaissance of SaaS and cloud data stores. This is where the heist happens. Unit 42 highlights systematic enumeration and theft from cloud platforms — CRM and case management systems, cloud storage (SharePoint/OneDrive, Google Workspace), data warehouses, and collaboration platforms. Attackers use the platforms' own native export and API capabilities: bulk report exports, list views scraped at API speed, full-object exports queued as "jobs." Because this is legitimate functionality used with valid credentials, traditional DLP and network controls frequently miss it.

5. RMM tooling for footholds that need it. Where actors want a durable endpoint presence (e.g., to reach on-prem file shares or harvest browser session tokens), they deploy legitimate remote monitoring and management tools — AnyDesk, TeamViewer, ScreenConnect-style agents — installed interactively or via scripted download. These tools are signed, widely whitelisted, and blend into environments where IT already uses one of them.

6. Extortion. Data is staged and exfiltrated, then the victim receives an extortion demand — often from an actor operating under a shifting persona — threatening public leak. Encryption is frequently not part of the playbook; pure data-theft extortion is faster, quieter, and harder to attribute.

Why Traditional Controls Fail

There is no exploit, no payload hash, no malicious domain at initial access. The failure points are procedural: weak caller verification at the help desk, self-service or agent-assisted MFA resets without step-up verification, permissive conditional access, and SaaS tenants with no anomaly detection on bulk export behavior. This is an identity and process attack, and it must be defended as one.

Detection & Response

The detections below target the highest-signal, lowest-noise behaviors in this chain: RMM tooling appearing where it shouldn't, MFA/identity changes clustered with help desk activity, and anomalous cloud export volumes.

Sigma Rules

YAML
---
title: Remote Access Tool Execution Outside Approved RMM
id: 9f1c2a84-3b7d-4e5f-9a10-2c8d7e6f5a4b
status: experimental
description: Detects execution of remote access tools commonly deployed by Com affiliates after vishing-based initial access (AnyDesk, TeamViewer, ScreenConnect, SplashTop, Level, FleetDeck). Baseline your sanctioned RMM first and exclude it.
references:
  - https://unit42.paloaltonetworks.com/com-affiliate-vishing-attacks/
  - https://attack.mitre.org/techniques/T1219/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.command_and_control
  - attack.t1219
logsource:
  category: process_creation
  product: windows
detection:
  selection_img:
    Image|endswith:
      - '\AnyDesk.exe'
      - '\TeamViewer.exe'
      - '\TeamViewer_Service.exe'
      - '\ScreenConnect.ClientService.exe'
      - '\strwinclt.exe'
      - '\level.exe'
      - '\fleetdeck_agent.exe'
      - '\fleetdeck_commander.exe'
  selection_download:
    CommandLine|contains:
      - 'anydesk.com/download'
      - 'download.teamviewer.com'
      - 'screenconnect.com'
  condition: 1 of selection_*
falsepositives:
  - Environments where one of these tools is the sanctioned RMM platform — tune to flag only non-standard tools
level: high
---
title: Suspicious MFA or Identity Reset Followed by Authenticator Registration
id: 2b8e4d61-7c3a-4f29-8e15-6d0a9c4b7f2e
status: experimental
description: Detects user risk-relevant identity changes in Microsoft Entra ID consistent with vishing-driven account takeover — authentication method registration or password reset events that should be correlated with help desk tickets.
references:
  - https://unit42.paloaltonetworks.com/com-affiliate-vishing-attacks/
  - https://attack.mitre.org/techniques/T1556/
  - https://attack.mitre.org/techniques/T1098/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.persistence
  - attack.t1098
  - attack.t1556
logsource:
  product: azure
  service: auditlogs
detection:
  selection:
    operationName|contains:
      - 'User registered security info'
      - 'User changed their password'
      - 'Reset user password'
      - 'Admin reset user password'
      - 'Add phone authentication method'
      - 'User started registration of security info'
  condition: selection
falsepositives:
  - Legitimate onboarding and help desk activity — alert only when not matched to a verified ticket, or when occurring outside business hours
level: medium
---
title: Cloud Data Export or Bulk Download Tooling Activity
id: 4c7a1f93-5e2b-48d6-b390-8f1e6a2c5d4a
status: experimental
description: Detects processes commonly used to stage and exfiltrate SaaS/cloud data after account compromise — WinRAR/7-Zip archiving of user profile cloud sync folders, and file transfer utilities observed in data-theft extortion intrusions.
references:
  - https://unit42.paloaltonetworks.com/com-affiliate-vishing-attacks/
  - https://attack.mitre.org/techniques/T1560/
  - https://attack.mitre.org/techniques/T1048/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.collection
  - attack.t1560
  - attack.exfiltration
  - attack.t1048
logsource:
  category: process_creation
  product: windows
detection:
  selection_archive:
    Image|endswith:
      - '\rar.exe'
      - '\7z.exe'
      - '\7za.exe'
    CommandLine|contains:
      - '\OneDrive\'
      - '\SharePoint\'
      - '\Google Drive\'
      - 'Documents'
  selection_transfer:
    Image|endswith:
      - '\rclone.exe'
      - '\FileZilla.exe'
      - '\winscp.exe'
      - '\megacmd.exe'
  condition: 1 of selection_*
falsepositives:
  - Legitimate archiving and managed file transfer — tune against approved software inventory and service accounts
level: medium

KQL — Microsoft Sentinel / Defender

The highest-fidelity hunt for this campaign is the identity change cluster: an MFA or password event followed quickly by a sign-in from new infrastructure. This query correlates Entra audit events with anomalous sign-ins within a short window.

KQL — Microsoft Sentinel / Defender
// Vishing-driven ATO: identity change followed by sign-in from new ASN/device within 60 minutes
let lookback = 14d;
let window = 60m;
let IdentityChanges =
    AuditLogs
    | where TimeGenerated > ago(lookback)
    | where OperationName has_any ("Reset user password", "Admin reset user password",
                                   "User registered security info", "Add phone authentication method",
                                   "Update user")
    | mv-expand TargetResources
    | extend TargetUser = tostring(TargetResources.userPrincipalName)
    | where isnotempty(TargetUser)
    | project ChangeTime=TimeGenerated, OperationName, TargetUser,
              InitiatedBy=tostring(parse_json(InitiatedBy).user.userPrincipalName),
              InitiatingIP=tostring(parse_json(InitiatedBy).user.ipAddress);
let SuspiciousSignins =
    SigninLogs
    | where TimeGenerated > ago(lookback)
    | where ResultType == 0
    | extend AuthDetail = tostring(parse_json(AuthenticationDetails)[0].authenticationMethod)
    | project SigninTime=TimeGenerated, UserPrincipalName, IPAddress, Location,
              AppDisplayName, DeviceDetail, AuthDetail;
IdentityChanges
| join kind=inner SuspiciousSignins on $left.TargetUser == $right.UserPrincipalName
| where SigninTime between (ChangeTime .. ChangeTime + window)
| extend TimeDelta = datetime_diff('minute', SigninTime, ChangeTime)
| project ChangeTime, OperationName, TargetUser, InitiatedBy, InitiatingIP,
          SigninTime, SigninIP=IPAddress, Location, AppDisplayName, AuthDetail, TimeDelta
| order by ChangeTime desc;

And a companion hunt for unsanctioned RMM across endpoints:

KQL — Microsoft Sentinel / Defender
// Hunt for RMM tooling execution outside approved software baseline
let ApprovedRMM = dynamic(["TeamViewer"]); // replace with your sanctioned tool(s), or empty to see all
DeviceProcessEvents
| where TimeGenerated > ago(7d)
| where FileName in~ ("AnyDesk.exe", "TeamViewer.exe", "TeamViewer_Service.exe",
                      "ScreenConnect.ClientService.exe", "strwinclt.exe",
                      "level.exe", "fleetdeck_agent.exe", "rclone.exe",
                      "winscp.exe", "FileZilla.exe", "megacmd.exe")
| extend Tool = tostring(split(FileName, ".")[0])
| where Tool !in~ (ApprovedRMM)
| summarize FirstSeen=min(TimeGenerated), LastSeen=max(TimeGenerated),
            Devices=dcount(DeviceId), SampleCmdline=any(ProcessCommandLine),
            Accounts=make_set(AccountName)
  by FileName, FolderPath, SHA256
| order by Devices desc;

For SaaS export anomalies — if you ingest Salesforce, Okta, or cloud app audit telemetry into Sentinel (via CEF/Syslog or the relevant connectors) — hunt for export job spikes:

KQL — Microsoft Sentinel / Defender
// Anomalous bulk export activity in SaaS audit logs (adjust table/fields to your connector)
CommonSecurityLog
| where TimeGenerated > ago(7d)
| where Message has_any ("DataExport", "ReportExport", "BulkExport", "export")
| summarize ExportCount=count(), Objects=make_set(RequestURL, 20)
  by SourceUserID, SourceIP, bin(TimeGenerated, 1h)
| where ExportCount > 10   // tune to baseline; exports per user per hour
| order by ExportCount desc;

Velociraptor VQL

For endpoint triage on a suspected compromised host — enumerate RMM presence and recent installs:

VQL — Velociraptor
-- Hunt for unauthorized RMM tools and their install artifacts across endpoints
SELECT Pid, Name, Exe, CommandLine, Username, CreateTime
FROM pslist()
WHERE Exe =~ '(?i)(anydesk|teamviewer|screenconnect|strwinclt|fleetdeck|level\.exe|rclone|megacmd|winscp|filezilla)'
VQL — Velociraptor
-- Enumerate RMM-related files in common staging locations (last 30 days)
SELECT FullPath, Size, Mtime, Ctime
FROM glob(globs=[
  'C:/Users/*/Downloads/*.exe',
  'C:/ProgramData/*/*.exe',
  'C:/Users/Public/*.exe'
])
WHERE FullPath =~ '(?i)(anydesk|teamviewer|screenconnect|connectwise|splashtop|fleetdeck|rclone|megacmd)'
  AND Mtime > (now() - 2592000)
ORDER BY Mtime DESC

Remediation / Hardening Script

The single most important technical countermeasure is eliminating SMS/voice as MFA factors for privileged and targeted populations and enforcing phishing-resistant MFA. This PowerShell audits your Entra ID tenant for users who still have phone-based methods registered and flags recent authentication-method changes that warrant help-desk ticket reconciliation.

PowerShell
#Requires -Modules Microsoft.Graph.Authentication, Microsoft.Graph.Users, Microsoft.Graph.Identity.SignIns
# Audit Entra ID for weak MFA methods and recent auth-method changes (vishing ATO indicators)
# Run with an account holding User.Read.All, AuditLog.Read.All, UserAuthenticationMethod.Read.All

Connect-MgGraph -Scopes "User.Read.All","AuditLog.Read.All","UserAuthenticationMethod.Read.All" -NoWelcome

$report = @()

# 1. Find users still using SMS/voice phone methods (phishable factors)
$users = Get-MgUser -All -Property "id,userPrincipalName,displayName" -Filter "accountEnabled eq true"
foreach ($u in $users) {
    try {
        $phone = Get-MgUserAuthenticationPhoneMethod -UserId $u.Id -ErrorAction SilentlyContinue
        if ($phone) {
            $report += [pscustomobject]@{
                Finding    = 'Phishable MFA factor (SMS/Voice) registered'
                User       = $u.UserPrincipalName
                Detail     = ($phone.PhoneType -join ',')
                Risk       = 'High'
                Action     = 'Migrate to FIDO2/passkey or certificate-based auth; block SMS via Authentication Methods policy'
            }
        }
    } catch { continue }
}

# 2. Authentication method changes in the last 7 days — reconcile every one against a help desk ticket
$start = (Get-Date).AddDays(-7).ToString('yyyy-MM-ddTHH:mm:ssZ')
$audits = Get-MgAuditLogDirectoryAudit -All -Filter "activityDateTime ge $start" |
    Where-Object { $_.ActivityDisplayName -match 'security info|password|authentication method' }
foreach ($a in $audits) {
    $target = ($a.TargetResources | Select-Object -First 1).UserPrincipalName
    $report += [pscustomobject]@{
        Finding    = "Identity change: $($a.ActivityDisplayName)"
        User       = $target
        Detail     = "Initiated by: $(($a.InitiatedBy.User.UserPrincipalName)) at $($a.ActivityDateTime)"
        Risk       = 'Review'
        Action     = 'Match to verified help desk ticket with out-of-band callback; if none exists, treat as ATO and revoke sessions'
    }
}

$report | Sort-Object Risk, User | Export-Csv -Path ".\EntraID-VishingExposureAudit-$(Get-Date -Format yyyyMMdd).csv" -NoTypeInformation
$report | Format-Table -AutoSize
Write-Host "`nDone. Review findings, then enforce: Authentication Methods policy (disable SMS/Voice), Conditional Access requiring phishing-resistant MFA for admins, and token protection CA policies." -ForegroundColor Yellow

Remediation

There is no patch for a phone call. Remediation here is architectural and procedural — prioritized by impact:

1. Re-engineer Help Desk Identity Verification (Highest Priority)

  • Implement mandatory step-up verification for all password resets and MFA changes: out-of-band callback to a number on file registered before the request, manager confirmation for privileged users, or live video verification against a government ID for high-risk roles.
  • Prohibit MFA resets based solely on knowledge-based answers (SSN, DOB, employee ID) — all of this is purchasable from data brokers and is exactly what Com operators arm themselves with.
  • Ticket everything. Any identity change without a matching ticket ID should be technically impossible for the agent to execute, and any change that does occur should trigger an automated SOC alert (see KQL above).
  • Publish an internal number employees can call to verify that "IT" actually contacted them — and train them to use it.

2. Eliminate Phishable Authentication Factors

  • Disable SMS and voice call as MFA methods tenant-wide via Entra ID Authentication Methods policy or Okta authenticator enrollment policies. These are the factors vishing and SIM swapping defeat.
  • Require phishing-resistant MFA (FIDO2/passkeys, Windows Hello for Business, certificate-based authentication) for all administrators, help desk staff, finance, and executives first — then the general population.
  • Enforce Conditional Access token protection and require compliant/hybrid-joined devices for access to sensitive SaaS (EHR-adjacent systems, CRM, financial platforms), so a stolen session cookie or fresh-factor login from attacker infrastructure fails.
  • Alert on and investigate new MFA method registrations, especially when followed by sign-ins from new ASNs, VPN egress, or impossible geographies within the hour.

3. Detect and Constrain Cloud Data Theft

  • Enable and baseline bulk export auditing in every business-critical SaaS platform (CRM report exports, data warehouse queries, SharePoint/OneDrive mass downloads). Alert on deviation from per-user baselines, not static thresholds.
  • Restrict export capabilities by role — most users have no business running full-object exports.
  • Deploy CASB/SSE controls to flag downloads to unmanaged devices and anomalous session volumes.

4. Control the RMM Layer

  • Maintain an explicit allowlist of sanctioned remote access tooling; alert on (or block via AppLocker/WDAC) everything else. Com affiliates specifically choose RMM tools you already trust or haven't banned.
  • Block egress to known RMM and exfiltration service domains where there is no business need.

5. Prepare for Extortion-Specific Response

  • Update IR playbooks: pure data-theft extortion without encryption requires different decisions — legal/regulatory notification analysis (HIPAA Breach Notification Rule, GLBA, state privacy laws), evidence preservation of SaaS audit logs before retention windows close, and leak-site monitoring.
  • Verify your SaaS audit log retention now. Salesforce, Entra ID, and Okta default retention periods are shorter than most investigations need. Export and archive.
  • Exercise the scenario. Run a tabletop where the entry point is a help desk agent, not a phishing email — most organizations discover their playbooks assume malware.

Review Unit 42's full write-up for the campaign-specific indicators and actor infrastructure they track, and fold those IOCs into your detections alongside the behavioral content above: https://unit42.paloaltonetworks.com/com-affiliate-vishing-attacks/

The Bottom Line

The Com's affiliates are succeeding because they attack the one control most organizations never hardened: the human workflow around identity recovery. Every control that matters here — help desk verification rigor, phishing-resistant MFA, export anomaly detection, RMM governance — is within your reach this quarter. The organizations getting extorted aren't the ones with unpatched VPNs this time. They're the ones whose help desk was polite, helpful, and unverified.

Related Resources

Security Arsenal Healthcare Cybersecurity AlertMonitor Platform Book a SOC Assessment healthcare Intel Hub

Is your security operations ready?

Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.