Skip to main content
ACTIVELY MONITORED//Verified: Nov 16, 2026
Threat IntelligenceAI SecurityMalware & Botnets
5 min read

FaceStealer: How to Detect DPRK $400M Fake Video Call Attacks

••By Rutvij Khatri•VERIFIED INTEL
FaceStealer: How to Detect DPRK $400M Fake Video Call Attacks

SOC Briefing Summary :: Executive Key Takeaways

  • [01]Threat summary: North Korean state actor BlueNoroff (APT38) steals over $400M in cryptocurrency by luring crypto executives and developers into rigged video calls.
  • [02]Root cause vector: Account takeover of trusted contacts combined with prerecorded stolen victim webcam footage, generative AI deepfakes, and ClickFix lures.
  • [03]Immediate action: Enforce out-of-band video call verification, block lookalike meeting domains, and restrict script execution from unmanaged browser contexts.

Executive Summary

The U.S. Federal Bureau of Investigation (FBI) and the Department of Justice (DoJ) have released an urgent cybersecurity advisory exposing FaceStealer, a highly deceptive, multi-stage cyber espionage and financial theft operation orchestrated by North Korean state-sponsored threat actors. Attributed to the prolific state-aligned cybercrime syndicate BlueNoroff (also tracked as APT38, Stardust Chollima, and Sapphire Sleet), the campaign has plundered more than $400 million in cryptocurrency over the past 18 months to fund Pyongyang's illicit ballistic missile and weapons programs.

The operation weaponizes a self-propagating social engineering cycle: threat actors compromise trusted corporate communication accounts (Telegram, LinkedIn, Slack, Discord), lure high-value targets into rigged videoconferences, stream stolen prerecorded webcam video or generative AI deepfakes of people the victim personally knows, and trick the target into executing malicious ClickFix terminal scripts. The FBI revealed that it has seized the attackers' cloud infrastructure, recovering more than 900 stolen victim webcam video recordings that were actively being used to deceive the next round of targets.

---

Technical Vulnerability Analysis & Attack Chain

Attack Chain Flow
// Attack Chain Flow

The FaceStealer attack chain represents a major evolution in social engineering, seamlessly blending account takeover, synthetic media deception, and client-side terminal exploitation:

1. Trusted Account Takeover & Conversational Ingress

The campaign begins not with cold phishing, but through hijacked legitimate communications accounts belonging to prior victims. BlueNoroff operators access compromised Telegram, LinkedIn, Slack, and Discord accounts, study past conversational history, and message the victim's professional contacts. Invitations are delivered via legitimate scheduling platforms (such as Calendly), proposing informal catch-up calls or investment discussions.

2. Synthetic Meeting Room & Stolen Webcam Streaming

Minutes before the scheduled call, the attacker delivers a link pointing to an adversary-controlled lookalike domain (e.g., uu01webzoom[.]us). When the victim joins:

  • The browser renders a high-fidelity replica of Zoom or Microsoft Teams.
  • Instead of a live presenter, the platform streams a prerecorded webcam recording of a prior victim known to the target, or a high-quality generative AI avatar mimicking a prominent cryptocurrency founder.
  • Simultaneously, the meeting application requests camera permissions and surreptitiously records the new victim's video feed, harvesting high-resolution facial footage to fuel future social engineering campaigns against the new victim's contacts.

3. ClickFix Audio Lure & Client-Side Detonation

After a few seconds of initial video interaction, the synthetic meeting room displays an alert stating that an audio glitch has occurred (e.g., "Audio driver error: Unable to detect microphone stream. Click below to apply fix"):

  • The victim is directed to copy a shell command string into their operating system terminal (PowerShell on Windows, Terminal on macOS).
  • The command executes a disguised base64-encoded one-liner that fetches cross-platform malware—frequently variants of the BeaverTail loader and InvisibleFerret Python backdoor.

4. Wallet Exfiltration & Software Supply Chain Pivoting

Once established, the malware scans browser local storage, cryptocurrency wallet extensions (MetaMask, Phantom, Coinbase Wallet), and memory for seed phrases and private keys. BlueNoroff operators frequently exhibit patient operational dwell times—waiting months for wallet balances to grow before transferring digital assets. In several high-profile cases, the operators used access to open-source software maintainers' devices to attempt unauthorized package updates on package registries like npm and crates.io.

---

MITRE ATT&CK Tactics, Techniques & Procedures (TTPs)

MITRE ATT&CK • OPERATIONAL TTP MAPPING
TacticTechnique IDTechnique NameOperational Context
Initial AccessT1566.002Spearphishing LinkLures sent via compromised Telegram, LinkedIn, and Slack accounts of trusted contacts.
ExecutionT1204.001User Execution: Malicious LinkVictim joins fraudulent video meeting room hosted on lookalike domain.
Defense EvasionT1656ImpersonationStreaming stolen victim webcam recordings and AI deepfakes to establish false trust.
ExecutionT1059.001PowerShellCoercing user into pasting base64-encoded ClickFix repair command into PowerShell.
ExecutionT1059.006PythonExecution of multi-stage Python backdoor (InvisibleFerret) in memory.
Credential AccessT1555.003Credentials from Web BrowsersHarvesting cryptocurrency wallet private keys, extension state, and session tokens.
Lateral MovementT1534Internal SpearphishingReusing compromised Slack and Discord credentials to pivot into the victim's organization.
ImpactT1496Financial TheftExfiltration and laundering of cryptocurrency assets to DPRK state-controlled wallets.

---

Threat Actor Profile & Exploitation Telemetry

  • Attributed Syndicate: BlueNoroff (Sub-cluster of the Lazarus Group / Reconnaissance General Bureau of the DPRK). Also tracked as APT38, Sapphire Sleet, Stardust Chollima, CageyChameleon, and UNC1069.
  • Targeting Profile: 80%+ of targets are C-suite executives, founders, and core engineers in cryptocurrency, blockchain finance, and high-impact open-source software repositories (including Axios and major package libraries).
  • Seized Infrastructure: FBI recovery of over 900 victim webcam videos confirms a cyclical, self-sustaining victim acquisition model operating across 58 countries.
  • AI Tooling Weaponization: Threat actors scraped YouTube interviews and conference panels, using commercial generative AI models (ChatGPT, Grok, Wondershare AIGC) to generate synthetic voice and video artifacts.

---

Detection & SOC Mitigation Playbook

1. Verification & Procedural Defenses

  • [ ] Enforce Webcam Safety Rules: Never enable corporate webcams until bidirectional real-time audio and video conversation has been verified with the host.
  • [ ] Mandate Out-of-Band Meeting Verification: If a meeting invite arrives unexpectedly over Telegram, Slack, or LinkedIn, confirm the meeting via phone call or direct corporate email.
  • [ ] Launch via Native Desktop Apps: Join video conferences exclusively by manually entering meeting IDs into official, locally installed Zoom or Teams desktop clients rather than clicking web browser links.
  • [ ] Prohibit Terminal Paste Actions: Train personnel never to copy-paste command strings from web meeting prompts or troubleshooting dialogs into terminal prompts.

2. Sigma Detection Rule (ClickFix Browser-Spawned PowerShell Execution)

QUERY / DETECTION_RULE
SIGMA / YAML
title: Browser Spawned PowerShell ClickFix Execution (FaceStealer)
id: 8e912401-419b-4e12-b901-3f9211d0a479
status: stable
description: Detects web browsers (Chrome, Edge, Brave, Firefox) spawning PowerShell, cmd, or bash processes to execute encoded commands or external web downloads
references:
  - https://thehackernews.com/2026/11/fbi-says-north-korean-hackers-stole.html
author: CyberNewsAI Threat Research Team
date: 2026-11-13
logsource:
  category: process_creation
  product: windows
detection:
  selection_parent:
    ParentImage|endswith:
      - '\chrome.exe'
      - '\msedge.exe'
      - '\brave.exe'
      - '\firefox.exe'
  selection_child:
    Image|endswith:
      - '\powershell.exe'
      - '\cmd.exe'
      - '\pwsh.exe'
  selection_flags:
    CommandLine|contains:
      - '-enc'
      - '-encodedcommand'
      - 'downloadstring'
      - 'invoke-webrequest'
      - 'curl'
      - 'iex'
  condition: selection_parent and selection_child and selection_flags
falsepositives:
  - Highly unusual developer workflows
level: high
tags:
  - attack.execution
  - attack.t1059.001
  - attack.initial_access
  - attack.t1566.002

3. Microsoft Sentinel & Defender Hunting Query (KQL)

QUERY / DETECTION_RULE
SENTINEL / KQL
// Hunt for processes initiated from clipboard or browser execution following connections to lookalike meeting domains
DeviceProcessEvents
| where InitiatingProcessFileName in~ ("chrome.exe", "msedge.exe", "brave.exe", "firefox.exe")
| where FileName in~ ("powershell.exe", "cmd.exe", "bash.exe", "sh.exe", "python.exe")
| where ProcessCommandLine has_any ("IEX", "DownloadString", "curl", "wget", "base64", "-enc", "http")
| project Timestamp, DeviceName, InitiatingProcessFileName, FileName, ProcessCommandLine, AccountName
| order by Timestamp desc

---

Indicator TypeValueContext / Telemetry
Phishing Domainuu01webzoom[.]usFraudulent Zoom meeting lure domain
Phishing Domainzoom-us-web[.]comImpersonation video infrastructure
Malware FamilyBeaverTailJavaScript infostealer and loader
Malware FamilyInvisibleFerretMulti-stage Python backdoor

---

Related Topic Cluster Intelligence

Indicators of Compromise (IOCs)
4 Identified
DOMAINuu01webzoom[.]us
DOMAINzoom-us-web[.]com
MALWARE FAMILYBeaverTail
MALWARE FAMILYInvisibleFerret

// VERIFIED_SOURCES_&_REFERENCES

SHARE INTEL:Reddit