Back to Intelligence

StealC RAT Resurrection (Stadia), Armorsteal SEO Poisoning & LLMjacking: OTX Pulse Analysis — Infostealer & Cloud Credential Theft Detection Pack

SA
Security Arsenal Team
November 21, 2026
9 min read

This week's OTX pulses reveal a coordinated surge in credential theft and monetization infrastructure spanning endpoints, browsers, and cloud control planes. Four distinct campaigns share a single objective: harvesting identities and converting them into revenue.

1. StealC returns as "Stadia" RAT. Four months after Operation Endgame dismantled StealC's infrastructure in June, the same seller re-emerged in October with Stadia — a full remote access trojan that reuses StealC's infostealer code verbatim. Stadia adds hidden VNC (HVNC) desktop control, a live command shell, file explorer, process manager, and an in-memory loader. The criminal ecosystem absorbed a major law-enforcement takedown and regenerated in under 120 days.

2. Armorsteal via SEO poisoning. A September campaign poisoned search results with typosquatted domains impersonating Chocolatey and Microsoft Activation Scripts (MAS), including massgravel.net. Victims searching for legitimate install commands executed malicious PowerShell that deployed a three-component Python-based stealer protected by Pyarmor, including remote access, clipboard hijacking, and credential harvesting modules.

3. One phishing kit, 29 fake banks. A single template kit spawned 29 professional fake-banking domains (brayfinance.online, qtbnk.com, oakmontfb.com, allyunitybn.com, etc.) targeting US and Canadian finance customers. Shared artifacts — identical taglines, routing numbers, and support phone numbers — confirm a common operator harvesting banking credentials and PII at scale.

4. Phishing-to-LLMjacking. Threat actors are using sponsored search ads to phish Google Cloud Platform privileged users, then manipulating organizational policies to disable security guardrails, minting service accounts with Gemini API access, and inviting external attacker accounts — hijacking victim billing to run LLM workloads. Typosquat domains such as groosl.com, groonl.com, and gcoalg.com imitate Google properties.

The common thread: adversaries are collapsing the distance between credential theft and monetization. Stolen browser cookies, banking logins, and cloud identities are all being converted — via RAT resale, crypto clipboard swaps, PII fraud, or cloud compute abuse — within days of collection.

Threat Actor / Malware Profile

Stadia (StealC successor)

  • Distribution: Re-released by the original StealC seller on criminal forums post-Operation Endgame; sold as MaaS.
  • Payload behavior: Verbatim reuse of StealC's stealer core (browser credentials, cookies, crypto wallets, messaging tokens) wrapped in a full RAT: hidden desktop (HVNC), interactive shell, file explorer, process manager.
  • C2: Communicates with panel infrastructure on domains such as stadiadev.cc.
  • Evasion: In-memory loader reflects later-stage payloads, avoiding on-disk artifacts; no persistent files required for the loader stage.

Armorsteal

  • Distribution: SEO poisoning and typosquatting against software-install search queries (Chocolatey, MAS). Malicious PowerShell one-liners executed by users.
  • Payload behavior: Three-component Python stealer — remote access module, clipboard hijacker (crypto address substitution), and credential harvester.
  • Evasion: PyArmor obfuscation defeats static analysis and casual script inspection; Python runtime dropped to disk masquerades as legitimate tooling.

Fake-Bank Phishing Kit

  • Distribution: Disposable domains (.online, .live, .cc, .com) registered in bulk; 29 unique bank brands from one template.
  • Behavior: Cloned login flows capture credentials plus full PII; identical routing number, tagline, and phone number across all 29 sites enable template-based detection.

LLMjacking Operation

  • Distribution: Sponsored search ads → fake Google login pages on typosquats (groosl.com, gbrokl.space, gcoalg.com).
  • Post-compromise: Organization-policy manipulation to disable guardrails, service-account creation with Gemini API scope, cross-tenant invitations to attacker accounts, and billing-destination abuse — monetizing stolen identities as AI compute resale.

IOC Analysis

The pulse set contains three operational indicator classes:

  1. Domains / hostnames (46+): C2 (stadiadev.cc), delivery typosquats (massgravel.net, groosl.com, gbrokl.xyz), and phishing infrastructure (qtbnk.com, allyunitybn.com, bank.firstcp.live). These are the highest-fidelity, lowest-latency indicators — push to DNS sinkholes, secure web gateway blocklists, and mail gateway URL rewriting immediately. Typosquat patterns (repeated o, transposed letters on google) should be converted into regex-based detection, not just exact-match blocks.
  2. File hashes (MD5/SHA1/SHA256): StealC/Stadia binaries and Armorsteal components (e.g., SHA256 0e5337a4cc8ebc772bc50f6e79befd1e8efad3410022e050d08d5269733c4ea4, ea928494fd30d88b98141d39a502a4324889fb732e74b44bfee12c72137dd73e). Ingest into EDR custom indicator lists and DeviceFileEvents retro-hunts. Treat hashes as short-lived — the Stadia in-memory loader means hash coverage alone will miss in-flight infections.
  3. Behavioral anchors: PyArmor-packed Python interpreters spawned by PowerShell, HVNC-style hidden desktop sessions, GCP org-policy changes followed by new service accounts — these outlive the IOCs and should drive detection engineering.

Operationalization: Export pulses via the OTX DirectConnect API into your SIEM/TIP; MISP or OpenCTI decode OTX natively. Prioritize domain indicators for network controls and use hashes for retrospective hunting rather than forward-looking prevention.

Detection Engineering

YAML
---
title: Stadia RAT / StealC C2 Domain Resolution
id: 7a1f3c2e-4b5d-4e6a-9c01-1d2e3f4a5b01
status: experimental
description: Detects DNS resolution or network connection to known Stadia RAT (StealC successor) C2 infrastructure observed in OTX pulses after Operation Endgame re-emergence.
author: Security Arsenal Threat Intel
references:
    - https://www.varonis.com/blog/stealc-returns
date: 2026/11/21
modified: 2026/11/21
tags:
    - attack.command_and_control
    - attack.t1071
    - attack.t1568
logsource:
    category: dns
selection:
    query|contains:
        - 'stadiadev.cc'
condition: selection
falsepositives:
    - Threat intel sandbox detonation
level: high
---
title: Armorsteal SEO-Poisoning PowerShell Downloader Execution
id: 8b2e4d3f-5c6e-4f7b-8d12-2e3f4a5b6c02
status: experimental
description: Detects suspicious PowerShell download-and-execute patterns consistent with Armorsteal delivery via SEO-poisoned fake Chocolatey / MAS install pages, including spawning of Python interpreters.
author: Security Arsenal Threat Intel
references:
    - https://www.zscaler.com/blogs/security-research/seo-poisoning-campaign-delivers-new-armorsteal-malware
date: 2026/11/21
modified: 2026/11/21
tags:
    - attack.execution
    - attack.t1059.001
    - attack.t1105
logsource:
    category: process_creation
    product: windows
selection_img:
    Image|endswith:
        - '\powershell.exe'
        - '\pwsh.exe'
selection_cmd:
    CommandLine|contains:
        - 'Invoke-Expression'
        - 'IEX'
        - 'DownloadString'
        - 'irm '
        - 'iwr '
selection_child:
    ParentImage|endswith:
        - '\chrome.exe'
        - '\msedge.exe'
        - '\firefox.exe'
condition: selection_img and selection_cmd and selection_child
falsepositives:
    - Legitimate software deployment tooling run from admin consoles
level: high
---
title: GCP Guardrail Weakening Followed by Service Account Creation (LLMjacking)
id: 9c3f5e4a-6d7f-4a8c-9e23-3f4a5b6c7d03
status: experimental
description: Detects Google Cloud organization policy modification or constraint disablement closely associated with new service account creation, consistent with LLMjacking campaigns that strip guardrails before minting Gemini API-enabled identities.
author: Security Arsenal Threat Intel
references:
    - https://www.wiz.io/blog/from-phishing-to-llmjacking
date: 2026/11/21
modified: 2026/11/21
tags:
    - attack.defense_evasion
    - attack.t1562
    - attack.persistence
    - attack.t1136
logsource:
    product: gcp
    service: audit
selection_policy:
    protoPayload.methodName|contains:
        - 'SetOrgPolicy'
        - 'orgpolicy.policy.set'
        - 'UpdatePolicy'
selection_sa:
    protoPayload.methodName|contains:
        - 'CreateServiceAccount'
        - 'CreateServiceAccountKey'
condition: selection_policy or selection_sa
falsepositives:
    - Legitimate org-policy administration during cloud onboarding
level: medium
KQL — Microsoft Sentinel / Defender
// Security Arsenal — Unified hunt: Stadia C2, Armorsteal delivery, fake-bank & LLMjacking infrastructure
let C2Domains = dynamic(["stadiadev.cc", "massgravel.net"]);
let PhishDomains = dynamic(["brayfinance.online","firstcp.live","securecapitalcorporationlimited.com","qtbnk.com","norexgo.com","1stggbinc.cc","oakmontfb.com","allyunitybn.com"]);
let LLMSquats = dynamic(["groosl.com","groonl.com","gboogl.com","gbrokl.space","gbrokl.xyz","gcoalg.com","gcslom.com","gpoosl.com"]);
let AllIOCs = array_concat(C2Domains, PhishDomains, LLMSquats);
let NetworkHits =
    DeviceNetworkEvents
    | where TimeGenerated > ago(7d)
    | where RemoteUrl has_any (AllIOCs)
    | project TimeGenerated, DeviceName, InitiatingProcessFileName, InitiatingProcessCommandLine, RemoteUrl, RemoteIP, ActionType;
let DnsHits =
    DeviceEvents
    | where TimeGenerated > ago(7d)
    | where ActionType == "DnsQueryResponse"
    | extend Query = tostring(AdditionalFields.DnsQueryString)
    | where Query has_any (AllIOCs)
    | project TimeGenerated, DeviceName, Query, InitiatingProcessFileName;
let PSExec =
    DeviceProcessEvents
    | where TimeGenerated > ago(7d)
    | where FileName in~ ("powershell.exe","pwsh.exe")
    | where ProcessCommandLine has_any ("IEX","Invoke-Expression","DownloadString","irm ","iwr ")
    | where InitiatingProcessFileName in~ ("chrome.exe","msedge.exe","firefox.exe")
    | project TimeGenerated, DeviceName, FileName, ProcessCommandLine, InitiatingProcessFileName;
union NetworkHits, DnsHits, PSExec
| sort by TimeGenerated desc
PowerShell
# Security Arsenal — StealC/Stadia + Armorsteal endpoint hunt script
# Run via EDR live response or Intune remediation across the fleet.

$results = @()

# 1. Known malicious file hashes (OTX pulse indicators)
$badHashes = @(
    '028afae8b5e9953e1cd84a003c42fd85',
    '7739a8d0d101ca82b93bbd0acdb6d9af',
    '7324066f5825c125c351ab7ccebad14d',
    'd0bbb176db499a05ab91c2bea826043d',
    'e4cedb7e809bb4335f25ddf16ec1f4c4',
    'f11a904e20addfec59b3265948b23bbf',
    'f4b53e8e87887af7dc9c8321a138410e'
)
$searchPaths = @("$env:TEMP","$env:LOCALAPPDATA","$env:APPDATA","$env:USERPROFILE\Downloads")
foreach ($p in $searchPaths) {
    if (Test-Path $p) {
        Get-ChildItem -Path $p -Recurse -File -Include *.exe,*.dll,*.py,*.pyc -ErrorAction SilentlyContinue |
            ForEach-Object {
                try {
                    $h = (Get-FileHash $_.FullName -Algorithm MD5 -ErrorAction Stop).Hash.ToLower()
                    if ($badHashes -contains $h) {
                        $results += [pscustomobject]@{Type='HashMatch'; Detail=$_.FullName; Hash=$h}
                    }
                } catch {}
            }
    }
}

# 2. Suspicious Python interpreters spawned in user-writable paths (Armorsteal PyArmor runtime)
Get-CimInstance Win32_Process -Filter "Name='python.exe' OR Name='pythonw.exe'" -ErrorAction SilentlyContinue |
    Where-Object { $_.ExecutablePath -match 'AppData|Temp|ProgramData' } |
    ForEach-Object { $results += [pscustomobject]@{Type='SuspiciousPython'; Detail="$($_.ExecutablePath) | CMD: $($_.CommandLine)"; Hash=''} }

# 3. Run-key persistence referencing stealer artifacts
$runKeys = @(
    'HKCU:\Software\Microsoft\Windows\CurrentVersion\Run',
    'HKLM:\Software\Microsoft\Windows\CurrentVersion\Run'
)
foreach ($k in $runKeys) {
    if (Test-Path $k) {
        Get-ItemProperty $k | Get-Member -MemberType NoteProperty | ForEach-Object {
            $val = (Get-ItemProperty $k).($_.Name)
            if ($val -match 'python|AppData|Temp' -and $val -match '\.exe|\.py|powershell') {
                $results += [pscustomobject]@{Type='RunKeyPersistence'; Detail="$k\$($_.Name) = $val"; Hash=''}
            }
        }
    }
}

# 4. Active connections to known C2 / phishing infrastructure
$badDomains = @('stadiadev.cc','massgravel.net','qtbnk.com','allyunitybn.com','groosl.com','gcoalg.com')
foreach ($d in $badDomains) {
    $resolved = Resolve-DnsName $d -ErrorAction SilentlyContinue | Where-Object {$_.IPAddress}
    foreach ($r in $resolved) {
        $conn = Get-NetTCPConnection -RemoteAddress $r.IPAddress -ErrorAction SilentlyContinue
        if ($conn) { $results += [pscustomobject]@{Type='ActiveC2Connection'; Detail="$d -> $($r.IPAddress) state=$($conn.State)"; Hash=''} }
    }
}

$results | Format-Table -AutoSize
if ($results.Count -gt 0) { Write-Warning "POTENTIAL COMPROMISE: $($results.Count) artifacts found. Isolate host and escalate." }
else { Write-Output "No StealC/Stadia or Armorsteal artifacts detected on $env:COMPUTERNAME." }

Response Priorities

Immediate (0–4 hours)

  • Block all 54 pulse indicators at DNS resolvers, secure web gateways, and EDR network controls. Deploy the typosquat regex detections for Google lookalikes (g[b/r/p][o/r]{2,}.*) rather than exact-match only.
  • Retro-hunt DeviceNetworkEvents and DNS logs for 30 days against stadiadev.cc and the 29 fake-bank domains — any hit indicates active credential exposure.
  • Ingest the 14 file hashes into EDR blocklists and run the PowerShell hunt script fleet-wide via live response.

Within 24 hours

  • Force credential resets and revoke active sessions for any user whose endpoint resolved phishing or C2 domains; infostealer logs trade within hours, so assume cookies and session tokens are already sold.
  • Audit GCP (and other cloud tenants) for org-policy changes, new service accounts, service-account key creation, and external IAM member invitations in the past 30 days; rotate keys and disable unauthorized Gemini API enablement.
  • Check IdP sign-in logs for sessions originating near visits to the fake-bank kit — banking phishers commonly pivot to corporate SSO reuse.

Within 1 week

  • Deploy application control (WDAC/AppLocker) blocking unsigned Python interpreters and PowerShell spawned from browsers — this breaks the Armorsteal chain at its chokepoint.
  • Enforce cloud billing alerts, Gemini API quota notifications, and conditional access with phishing-resistant MFA (FIDO2) for all privileged cloud roles.
  • Roll out browser isolation or link-rewriting for sponsored search results, and brief the service desk on the fake-bank kit's shared artifacts (identical routing number/phone) for rapid user-report triage.

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.