Back to Intelligence

ClickFix + Lumma Stealer via Finger LOLBin, PAN-OS CVE-2026-0257 Ransomware Exploitation, and AI-Assisted Cybercrime: OTX Pulse Analysis — Enterprise Detection Pack

SA
Security Arsenal Team
November 20, 2026
12 min read

Three pulses published to AlienVault OTX between 19–20 November 2026 collectively map a converging threat landscape: low-friction social engineering at the edge, authentication bypass on perimeter appliances, and AI-lowered attacker skill floors.

Pulse 1 — Amazon-Themed ClickFix Campaign: Operators purchased sponsored Google search placement for the query "amazon business" to redirect victims to a spoofed Amazon login page hosted on Google Sites. The page serves a fake CAPTCHA — the ClickFix technique — that tricks users into pasting and executing a malicious command invoking Windows' obsolete finger.exe utility as a living-off-the-land binary (LOLBins) for payload retrieval. The infection chain ultimately delivers Lumma Stealer (credential/browser/session theft), Doina (loader/dropper stage), and a NetSupport Client RAT for persistent remote access. Hosting and staging infrastructure includes compromised or actor-registered domains such as grillmasterchef.com, ashevilleroofingcontractor.com, and grillrecipespro.com.

Pulse 2 — PAN-OS CVE-2026-0257 Active Exploitation: Unit 42 reports active exploitation of an authentication bypass in Palo Alto Networks PAN-OS GlobalProtect portal and gateway components. Ransomware operators Spikey Scorpius (Qilin), Tarnished Scorpius (Inc), and the Settra group are weaponizing the flaw to establish unauthorized VPN sessions, then pivoting to virtualization-layer attacks (Brave Prince tooling, S0252) and ransomware deployment. Targeting spans ten verticals including education, finance, healthcare, government, energy, and transportation — opportunistic rather than selective.

Pulse 3 — BUGSTER AI-Assisted Intrusion Operations: A financially motivated actor using jailbroken AI assistants (Hermes Agent framework with DeepSeek V4) conducted reconnaissance, exploit script generation, and exfiltration against education and telecommunications targets in India, stealing 5,200+ personal records. Notably, the actor has limited native technical skill — the AI toolchain substitutes for it. This materially lowers the cost of entry-level intrusion and increases expected incident volume.

Collective assessment: The three pulses share a common thread — attack surface erosion via trusted channels (search ads, Google Sites, VPN concentrators) combined with commodity tooling. Defenders should treat perimeter VPN appliances and identity/session theft as the highest-priority exposure this cycle.

Threat Actor / Malware Profile

Lumma Stealer + Doina + NetSupport (ClickFix chain)

  • Distribution: Malvertising via sponsored Google search results; spoofed login pages on Google Sites; ClickFix fake-CAPTCHA social engineering delivering clipboard-executed commands.
  • Payload behavior: finger.exe is abused to retrieve second-stage content over the legacy Finger protocol (TCP/79). Doina acts as the staging loader, which decrypts and injects Lumma Stealer. Lumma harvests browser credentials, cookies, session tokens, crypto wallets, and 2FA artifacts. A NetSupport RAT client is deployed for hands-on-keyboard follow-on access.
  • C2 communication: Lumma uses HTTP/S POST exfiltration to rotating C2 domains, frequently hidden behind Cloudflare. NetSupport RAT communicates over TCP 443/80 with its own protocol, masquerading as legitimate remote-admin traffic.
  • Persistence: NetSupport installs as a service or registry Run key entry; Lumma is typically smash-and-grab but the RAT stage provides durable access.
  • Anti-analysis: LOLBin abuse (finger.exe) evades signature and reputation controls; payload staging across benign-seeming recipe/contractor domains defeats domain categorization; in-memory injection avoids disk artifacts.

Spikey Scorpius (Qilin) / Tarnished Scorpius (Inc) / Settra

  • Initial access: Authentication bypass on PAN-OS GlobalProtect portal/gateway (CVE-2026-0257), yielding unauthorized VPN sessions.
  • Post-exploitation: Brave Prince (S0252) tooling for virtualization-layer attacks — targeting ESXi/vCenter to encrypt entire datastores at scale. Qilin and Inc ransomware payloads deployed after lateral movement and data theft (double extortion).
  • Persistence: Creation of rogue local admin accounts on firewall appliances, webshells/droppers on management interfaces, scheduled tasks and service creation on Windows/ESXi hosts.
  • Anti-analysis: Log tampering on appliances, use of legitimate admin protocols (SMB, WinRM, SSH) for lateral movement to blend with administrative noise.

BUGSTER (AI-assisted actor)

  • Method: Jailbroken LLM assistants (Hermes Agent + DeepSeek V4) generate reconnaissance scripts, exploit code, and exfiltration logic. Despite low actor skill, operational tempo is high. Expect AI-generated tooling to show distinctive tells: verbose comments, generic variable names, inconsistent error handling — useful for code-attribution hunts.

IOC Analysis

The indicator set across these pulses is weighted toward domains and file hashes, with CVEs serving as the pivot for exposure management rather than direct detection.

IndicatorTypePulseOperationalization
grillmasterchef.comDomainClickFixBlock at DNS/proxy; hunt proxy logs for connections
ashevilleroofingcontractor.comDomainClickFixBlock; note pattern — legitimate small-business sites used as staging
grillrecipespro.comDomainClickFixBlock; retro-hunt last 90 days of DNS resolution
035d396b...17e8b7 (SHA-256)FileHashClickFixBlock in EDR; hunt for process creation by hash
2bb937c7...768e25 (MD5) / 923db163...2477 (SHA-1)FileHashClickFixSupplement SHA-256 blocking for legacy tooling
CVE-2026-0257CVEPAN-OSPatch immediately; audit GlobalProtect logs for pre-patch compromise
CVE-2026-109618 / CVE-2026-109619CVEClickFix / BUGSTERCross-reference vulnerability scanners; shared CVEs across pulses suggest common exploit tooling
CVE-2023-27532CVEPAN-OS pulseVeeam Backup flaw referenced in operator playbook — verify Veeam patch state as ransomware actors target backups

Operational guidance: Ingest domains and hashes into your threat intel platform (MISP, OpenCTI) with TLP:WHITE handling; push domains to DNS sinkhole/proxy blocklists; push hashes to EDR block policies. Note the CVE overlap (CVE-2026-109618/619 appearing in both the ClickFix and BUGSTER pulses) — this indicates shared exploit kits or AI-generated tooling converging on the same vulnerabilities, and should drive those CVEs to the top of patch prioritization.

Detection Engineering

YAML
---
title: ClickFix Finger Utility LOLBin Execution
id: 7a1c4e52-9f3d-4b1e-8c2a-1d5e6f7a8b9c
status: experimental
description: Detects execution of Windows finger.exe with suspicious arguments consistent with ClickFix social-engineering chains that abuse the Finger protocol to retrieve payloads (Lumma Stealer / Doina staging).
author: Security Arsenal Threat Intelligence
date: 2026/11/20
references:
    - https://www.huntress.com/blog/amazon-clickfix-finger
logsource:
    category: process_creation
    product: windows
detection:
    selection_img:
        Image|endswith: '\finger.exe'
    selection_args:
        CommandLine|contains:
            - '@'
            - 'http'
            - 'cmd'
            - 'powershell'
            - 'mshta'
            - '|'
    condition: selection_img and selection_args
falsepositives:
    - Virtually none; finger.exe is obsolete and has no legitimate enterprise use case in modern environments
level: high
tags:
    - attack.execution
    - attack.t1059
    - attack.t1105
    - attack.t1659
---
title: NetSupport RAT or Lumma Stealer Staging Infrastructure Communication
id: 2b8d5f63-0a4e-5c2f-9d3b-2e6f708b9c0d
status: experimental
description: Detects network connections to domains identified in the Amazon-themed ClickFix campaign delivering Lumma Stealer, Doina, and NetSupport RAT.
author: Security Arsenal Threat Intelligence
date: 2026/11/20
references:
    - https://www.huntress.com/blog/amazon-clickfix-finger
logsource:
    category: dns
    product: windows
detection:
    selection:
        query|contains:
            - 'grillmasterchef.com'
            - 'ashevilleroofingcontractor.com'
            - 'grillrecipespro.com'
    condition: selection
falsepositives:
    - Unlikely; these domains have no legitimate business purpose in this context
level: critical
tags:
    - attack.command_and_control
    - attack.t1071
    - attack.t1105
---
title: Rogue VPN Session or GlobalProtect Auth Bypass Artifacts (CVE-2026-0257)
id: 3c9e6074-1b5f-6d30-0e4c-3f7081c0d1e2
status: experimental
description: Detects suspicious process execution on Windows hosts immediately following GlobalProtect VPN session establishment, consistent with ransomware operator post-exploitation behavior (Qilin/Inc/Settra) after PAN-OS CVE-2026-0257 authentication bypass.
author: Security Arsenal Threat Intelligence
date: 2026/11/20
references:
    - https://unit42.paloaltonetworks.com/active-exploitation-of-pan-os-cve-2026-0257/
logsource:
    category: process_creation
    product: windows
detection:
    selection_parent:
        ParentImage|endswith:
            - '\PanGPA.exe'
            - '\PanGPS.exe'
    selection_suspicious_child:
        Image|endswith:
            - '\cmd.exe'
            - '\powershell.exe'
            - '\wmic.exe'
            - '\psexec.exe'
            - '\nltest.exe'
            - '\whoami.exe'
    condition: selection_parent and selection_suspicious_child
falsepositives:
    - Rare; GlobalProtect processes do not normally spawn administrative shells
level: high
tags:
    - attack.initial_access
    - attack.t1190
    - attack.t1133
KQL — Microsoft Sentinel / Defender
// Hunt: ClickFix / Lumma Stealer finger.exe LOLBin + staging domains + post-VPN recon (CVE-2026-0257)
// Lookback: 14 days | Microsoft Sentinel / Defender XDR
let ClickFixDomains = dynamic(["grillmasterchef.com", "ashevilleroofingcontractor.com", "grillrecipespro.com"]);
let StagingHashes = dynamic(["035d396b3a329f59d680dc36a0a639eb7fb306bea97e42f5f53293dce817e8b7",
                             "2bb937c7eac0843313ee402735768e25",
                             "923db163b89884e3f371cf322ee1a071515e2477"]);
let FingerExec =
    DeviceProcessEvents
    | where TimeGenerated > ago(14d)
    | where FileName =~ "finger.exe"
    | extend SuspiciousArgs = ProcessCommandLine has_any ("@", "http", "cmd", "powershell", "mshta", "|")
    | project TimeGenerated, DeviceName, AccountName, ProcessCommandLine, InitiatingProcessFileName, InitiatingProcessCommandLine, FingerExec=1;
let DomainHits =
    DeviceNetworkEvents
    | where TimeGenerated > ago(14d)
    | where RemoteUrl in~ (ClickFixDomains)
    | project TimeGenerated, DeviceName, InitiatingProcessFileName, RemoteUrl, RemoteIP, RemotePort, DomainHit=1;
let HashHits =
    DeviceProcessEvents
    | where TimeGenerated > ago(14d)
    | where SHA256 in~ (StagingHashes) or SHA1 in~ (StagingHashes) or MD5 in~ (StagingHashes)
    | project TimeGenerated, DeviceName, FileName, SHA256, ProcessCommandLine, HashHit=1;
let PostVpnRecon =
    DeviceProcessEvents
    | where TimeGenerated > ago(14d)
    | where InitiatingProcessFileName in~ ("PanGPA.exe", "PanGPS.exe")
    | where FileName in~ ("cmd.exe", "powershell.exe", "wmic.exe", "whoami.exe", "nltest.exe", "net.exe")
    | project TimeGenerated, DeviceName, InitiatingProcessFileName, FileName, ProcessCommandLine, PostVpnRecon=1;
union isfuzzy=true FingerExec, DomainHits, HashHits, PostVpnRecon
| sort by TimeGenerated asc
PowerShell
<#
.SYNOPSIS
    Security Arsenal IOC Hunt — ClickFix/Lumma + PAN-OS CVE-2026-0257 artifacts
.DESCRIPTION
    Checks hosts for: finger.exe execution remnants, NetSupport RAT persistence
    (services/run keys), ClickFix staging domain connections in DNS cache,
    staging payload hashes, and rogue VPN-era local admin accounts.
    Run via your RMM/EDR or PS remoting across the fleet. Outputs JSON per host.
#>

$Report = [ordered]@{
    Hostname             = $env:COMPUTERNAME
    ScanTime             = (Get-Date).ToUniversalTime().ToString("o")
    Findings             = @()
}

# --- 1. DNS cache hits for ClickFix staging domains ---
$BadDomains = @("grillmasterchef.com","ashevilleroofingcontractor.com","grillrecipespro.com")
$DnsHits = Get-DnsClientCache -ErrorAction SilentlyContinue | Where-Object {
    $n = $_.Entry; $BadDomains | Where-Object { $n -like "*$_*" }
}
if ($DnsHits) {
    $Report.Findings += @{ Check = "ClickFixStagingDomain-DNSCache"; Severity = "CRITICAL"
        Detail = ($DnsHits | Select-Object Entry, Data | ConvertTo-Json -Compress) }
}

# --- 2. Staging payload file hashes anywhere on common drop paths ---
$BadHashes = @("035d396b3a329f59d680dc36a0a639eb7fb306bea97e42f5f53293dce817e8b7")
$DropPaths = @("$env:TEMP","$env:LOCALAPPDATA\Temp","$env:APPDATA","C:\Users\Public")
foreach ($p in $DropPaths) {
    Get-ChildItem -Path $p -Recurse -File -ErrorAction SilentlyContinue | ForEach-Object {
        $h = (Get-FileHash $_.FullName -Algorithm SHA256 -ErrorAction SilentlyContinue).Hash
        if ($h -and ($BadHashes -contains $h.ToLower())) {
            $Report.Findings += @{ Check = "LummaStagingHash"; Severity = "CRITICAL"; Detail = $_.FullName }
        }
    }
}

# --- 3. NetSupport RAT persistence: services, run keys, install dirs ---
$SvcHits = Get-CimInstance Win32_Service -ErrorAction SilentlyContinue |
    Where-Object { $_.Name -match "client32|netsupport|PCICTL" -or $_.PathName -match "client32\.exe|NSM" }
if ($SvcHits) {
    $Report.Findings += @{ Check = "NetSupportService"; Severity = "HIGH"
        Detail = ($SvcHits | Select-Object Name, PathName, State | ConvertTo-Json -Compress) }
}
$RunKeys = @("HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Run",
             "HKCU:\SOFTWARE\Microsoft\Windows\CurrentVersion\Run")
foreach ($rk in $RunKeys) {
    $props = Get-ItemProperty -Path $rk -ErrorAction SilentlyContinue
    if ($props) {
        $props.PSObject.Properties | Where-Object { $_.Value -match "client32|netsupport|NSM|finger" } | ForEach-Object {
            $Report.Findings += @{ Check = "PersistenceRunKey"; Severity = "HIGH"; Detail = "$($_.Name) = $($_.Value)" }
        }
    }
}

# --- 4. Scheduled tasks invoking LOLBins or suspicious payloads ---
Get-ScheduledTask -ErrorAction SilentlyContinue | ForEach-Object {
    $actions = ($_.Actions | ForEach-Object { "$($_.Execute) $($_.Arguments)" }) -join "; "
    if ($actions -match "finger\.exe|mshta|powershell.*-enc|client32") {
        $Report.Findings += @{ Check = "SuspiciousScheduledTask"; Severity = "HIGH"; Detail = "$($_.TaskName): $actions" }
    }
}

# --- 5. Live/established connections on Finger (79) or NetSupport default ports ---
$Conns = Get-NetTCPConnection -State Established -ErrorAction SilentlyContinue |
    Where-Object { $_.RemotePort -in @(79, 5405, 4421) -and $_.RemoteAddress -notmatch "^(10\.|172\.(1[6-9]|2[0-9]|3[0-1])\.|192\.168\.|127\.)" }
if ($Conns) {
    $Report.Findings += @{ Check = "SuspiciousOutboundPort"; Severity = "HIGH"
        Detail = ($Conns | Select-Object LocalPort, RemoteAddress, RemotePort, OwningProcess | ConvertTo-Json -Compress) }
}

# --- 6. Recently created local admin accounts (post-VPN-bypass rogue accounts) ---
$Cutoff = (Get-Date).AddDays(-14)
Get-LocalUser -ErrorAction SilentlyContinue | Where-Object { $_.Enabled -and $_.LastLogon -gt $Cutoff } | ForEach-Object {
    $isAdmin = (Get-LocalGroupMember -Group "Administrators" -ErrorAction SilentlyContinue).Name -contains "$env:COMPUTERNAME\$($_.Name)"
    if ($isAdmin -and $_.Name -notin @("Administrator","DefaultAccount","WDAGUtilityAccount")) {
        $Report.Findings += @{ Check = "RecentLocalAdminLogon"; Severity = "MEDIUM"; Detail = "$($_.Name) last logon $($_.LastLogon)" }
    }
}

$Report | ConvertTo-Json -Depth 5

Response Priorities

Immediate (0–4 hours)

  • Block grillmasterchef.com, ashevilleroofingcontractor.com, grillrecipespro.com at DNS resolver, secure web gateway, and EDR network protection layers. Retro-hunt 90 days of DNS/proxy logs.
  • Block the Lumma staging hashes (MD5/SHA-1/SHA-256) in EDR block policies.
  • Alert on any finger.exe execution — there is no legitimate use case in modern enterprise environments; treat any hit as a confirmed incident.
  • Patch PAN-OS to the fixed release addressing CVE-2026-0257 on all GlobalProtect portals/gateways. If patching is delayed, restrict portal access by source IP and enable enhanced logging.
  • Audit GlobalProtect auth logs for sessions from anomalous geographies/ASNs predating the patch; assume compromise where anomalies exist.

24 hours

  • Identity verification is mandatory. Lumma Stealer harvests browser session tokens and cookies — assume all sessions on affected endpoints are compromised. Force password resets and revoke active sessions (Entra ID / Okta) for any user on an endpoint with finger.exe execution or staging-domain hits.
  • Invalidate refresh tokens and revoke OAuth grants for impacted users; session-token theft bypasses MFA.
  • Hunt for NetSupport RAT persistence (services, run keys) per the PowerShell script above; any hit triggers full host isolation and forensic imaging.
  • Review VPN-connected hosts for post-authentication recon (whoami, nltest, net group) within 15 minutes of session establishment — a hallmark of the Scorpius operator playbooks.
  • Verify Veeam Backup patch state (CVE-2023-27532 referenced in the operator playbook) — ransomware crews target backups before detonation.

1 week

  • Attack-surface hardening: If GlobalProtect is internet-exposed, migrate to a ZTNA posture or place the portal behind an IP allowlist/geofence. Subscribe to PAN-OS emergency-release advisories.
  • LOLBins policy: Deploy WDAC/AppLocker rules blocking finger.exe, and audit other retro-utilities (finger, tftp, rexec) that adversaries repurpose for payload retrieval.
  • ClickFix user awareness: Distribute guidance on fake-CAPTCHA paste-and-run lures. No legitimate site will ever ask a user to open Run/Terminal and paste a command. Consider browser policies that disable clipboard-write prompts from untrusted origins.
  • Virtualization layer: Audit ESXi/vCenter for unauthorized accounts and SSH enablement; enforce lockdown mode and MFA on vCenter — this is the Qilin/Inc endgame target.
  • AI-assisted threat planning: Update threat models to assume higher intrusion volume from low-skill actors; prioritize automated containment over manual triage for commodity incidents.

Related Resources

Security Arsenal Incident Response Managed SOC & MDR Services AlertMonitor Threat Detection From The Dark Side Intel Hub

Is your security operations ready?

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