Skip to navigation

Threat reportAPTTL-2026-6095

FaceStealer: North Korean (BlueNoroff) deepfake/ClickFix social engineering campaign behind $400M+ crypto theft disrupted by US

highACTIVE

The US government disrupted FaceStealer, a North Korean social engineering campaign that used hijacked accounts, recorded/deepfake video calls and ClickFix-style fake-error 'fixes' against crypto/Web3 workers and open-source developers, stealing more than $400 million over about 18 months. The US seized the campaign's cloud-based infrastructure; Arctic Wolf attributes the activity to BlueNoroff.

Key facts

IdentifiersTL-2026-6095
Also known asFaceStealer
SeverityHIGH
First disclosed
Affected productsMultiple Zoom, Microsoft Teams and Google Meet users (meetings impersonated; credentials targeted); Multiple Telegram, LinkedIn, Slack and Discord accounts (hijacked and used to send meeting invites); Apple / Microsoft / Linux macOS, Windows and Linux endpoints of crypto/Web3 workers and open-source developers
AttributionAPT38 (high confidence)
Nation-state nexusNorth Korea
MalwareFaceStealer, NovaStream, telegram
ToolsAdobe Premiere, FakeUpdates, GPT-4o, Google Generative AI, Grok Imagine, Wondershare AIGC
MITRE ATT&CK techniquesT1059.001 PowerShell, T1059.002 AppleScript, T1078 Valid Accounts, T1125 Video Capture, T1195.001 Compromise Software Dependencies and Development Tools, T1204.004 Malicious Copy and Paste, T1534 Internal Spearphishing, T1566.003 Spearphishing via Service, T1583.001 Domains, T1586.001 Social Media Accounts, T1657 Financial Theft, T1684.001 Impersonation
MotivationFINANCIAL
Target sectorscryptocurrency, web3, finance, technology, open-source software
Target regionsGlobal, North America, Europe, Asia
Detection rules9 (1 shown free)
Sources9 cited sources

Detection coverage

As of 2026-11-14, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-6095 across Splunk SPL, Microsoft KQL and Sigma, covering 9 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: FaceStealer ClickFix Payload Fetch - osascript, curl-pipe-shell, PowerShell Cradle (Exploitation)

Behaviour-first detection of the ClickFix fake-fix stage used by FaceStealer/BlueNoroff (TL-2026-6095): osascript do shell script downloads (fake Chrome PDF viewer Update Now), curl/wget piped to a shell on Windows/Linux/macOS, and powershell.exe download-and-execute cradles. Uses the CIM Endpoint.Processes data model so it works across Sysmon, EDR and Linux/macOS process telemetry.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1204.004, T1059.001, T1059.002
  • Data sources: splunk_cim_endpoint_processes, sysmon, edr_process_telemetry
  • Severity: high
  • Confidence: high

Splunk SPL

| tstats summariesonly=false count min(_time) as firstTime max(_time) as lastTime from datamodel=Endpoint.Processes where Processes.process_name IN ("osascript","curl","wget","powershell.exe","pwsh.exe","bash","sh","zsh") by Processes.dest Processes.user Processes.parent_process_name Processes.process_name Processes.process
| rename "Processes.*" as *
| eval cmd=lower(process)
```FaceStealer ClickFix: victim-run fake fix fetches a payload; behaviours = AppleScript shell download, curl/wget piped to a shell, PowerShell download-and-execute cradle```
| eval osa_fetch=if(process_name="osascript" AND like(cmd,"%do shell script%") AND match(cmd,"(curl|wget|http)"),1,0)
| eval pipe_shell=if(match(cmd,"(curl|wget)\s[^|]*https?://[^|]*\|\s*(sudo\s+)?(ba|z)?sh"),1,0)
| eval ps_cradle=if(match(process_name,"(?i)^(powershell|pwsh)\.exe$") AND match(cmd,"(iwr|invoke-webrequest|irm\s|invoke-restmethod|downloadstring|downloadfile)") AND match(cmd,"(iex|invoke-expression|-enc|-encodedcommand|-w(indowstyle)?\s+hidden)"),1,0)
| where osa_fetch=1 OR pipe_shell=1 OR ps_cradle=1
| eval technique=case(osa_fetch=1,"T1059.002",ps_cradle=1,"T1059.001",true(),"T1204.004")
| convert ctime(firstTime) ctime(lastTime)
| table firstTime lastTime dest user parent_process_name process_name process technique

Microsoft KQL

DeviceProcessEvents
| where Timestamp > ago(24h)
| where FileName in~ ("osascript","curl","wget","powershell.exe","pwsh.exe","bash","sh","zsh")
| extend cmd = tolower(ProcessCommandLine)
| extend OsaFetch = FileName =~ "osascript" and cmd has "do shell script" and cmd has_any ("curl","wget","http")
| extend PipeShell = cmd matches regex @"(curl|wget)\s[^|]*https?://[^|]*\|\s*(sudo\s+)?(ba|z)?sh"
| extend PsCradle = FileName in~ ("powershell.exe","pwsh.exe")
    and cmd has_any ("iwr","invoke-webrequest","irm ","invoke-restmethod","downloadstring","downloadfile")
    and cmd has_any ("iex","invoke-expression","-enc","-encodedcommand","-w hidden","-windowstyle hidden")
| where OsaFetch or PipeShell or PsCradle
| project Timestamp, DeviceName, AccountName, InitiatingProcessFileName, InitiatingProcessCommandLine, FileName, ProcessCommandLine, OsaFetch, PipeShell, PsCradle

Sigma

title: FaceStealer ClickFix Remote Payload Fetch via osascript, Shell Pipe or PowerShell Cradle
id: 6095a000-0001-4000-8000-000000000001
status: experimental
description: Detects victim-executed ClickFix style fixes used by the FaceStealer / BlueNoroff campaign (TL-2026-6095) - osascript do shell script downloads, curl/wget piped to a shell, and PowerShell download-and-execute cradles.
references:
    - https://www.justice.gov/opa/pr/justice-department-announces-actions-disrupt-north-koreas-massive-theft-cryptocurrency
    - http://arcticwolf.com/resources/blog/still-in-the-meeting-bluenoroff-six-months-on/
    - https://www.bleepingcomputer.com/news/security/us-disrupts-north-korean-campaign-behind-400-million-crypto-theft/
author: Threadlinqs Team
date: 2026-11-14
tags:
    - attack.execution
    - attack.t1204.004
    - attack.t1059.001
    - attack.t1059.002
    - attack.defense-evasion
    - attack.t1656
logsource:
    category: process_creation
detection:
    selection_osascript:
        Image|endswith: '/osascript'
        CommandLine|contains|all:
            - 'do shell script'
            - 'http'
    selection_pipe_shell:
        Image|endswith:
            - '/bash'
            - '/sh'
            - '/zsh'
        CommandLine|re: '(curl|wget)\s[^|]*https?://[^|]*\|\s*(sudo\s+)?(ba|z)?sh'
    selection_ps_img:
        Image|endswith:
            - '\powershell.exe'
            - '\pwsh.exe'
    selection_ps_download:
        CommandLine|contains:
            - 'iwr '
            - 'Invoke-WebRequest'
            - 'irm '
            - 'Invoke-RestMethod'
            - 'DownloadString'
            - 'DownloadFile'
    selection_ps_exec:
        CommandLine|contains:
            - 'iex'
            - 'Invoke-Expression'
            - '-enc'
            - '-w hidden'
            - '-WindowStyle Hidden'
    condition: selection_osascript or selection_pipe_shell or (selection_ps_img and selection_ps_download and selection_ps_exec)
falsepositives:
    - Developer install scripts such as curl https://... | sh for Homebrew, rustup or nvm
    - Administrator PowerShell deployment one-liners
level: high

Known false positives:

  • Developer or admin bootstrap scripts using curl-pipe-shell from trusted vendors (Homebrew, rustup, nvm)
  • Software deployment tools invoking PowerShell download cradles

Tuning notes: Allowlist known installer hosts by extending the match with NOT match(cmd,"(brew.sh|rustup.rs|nodejs.org)"); scope to crypto/Web3 and developer hosts first; correlate with a Zoom, Teams or Google Meet process start within 10 minutes to raise fidelity.

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)
  • FaceStealer Windows ClickFix Fake Chrome Update Execution (Exploitation) · unlock with TL Creds
  • Package Manager or Build Script Spawning Remote Download-and-Execute (Supply Chain) · unlock with TL Creds
  • FaceStealer Typosquatted Meeting Link from Hijacked Account (Lateral Movement / Initial Access) · unlock with TL Creds
  • FaceStealer macOS osascript Downloader and Password Prompt (Exploitation) · unlock with TL Creds
  • FaceStealer Webcam Capture via CLI Tooling (Post-Exploitation Collection) · unlock with TL Creds
  • FaceStealer Lookalike Meeting Domain DNS Query (C2 / Delivery Infrastructure) · unlock with TL Creds
  • FaceStealer Credential and Crypto Wallet Data Staging (Post-Exploitation) · unlock with TL Creds
  • FaceStealer Community-Observed DPRK Sibling Infrastructure (C2 / Resource Development) · unlock with TL Creds

Indicators of compromise

As of 2026-11-14, Threadlinqs Intelligence tracks 9 indicators of compromise for TL-2026-6095.

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

Technical analysis

FaceStealer is the FBI's name for a North Korean social engineering scheme that ran for roughly 18 months and, per the US Department of Justice, stole more than $400 million in cryptocurrency. The DOJ (Central District of California, with the FBI Los Angeles Field Office) announced on 2026-11-12 that it had seized the scheme's cloud-based infrastructure, and the FBI seized more than 900 secretly recorded webcam videos of victims under a federal warrant. The DOJ lists the tracking names Sapphire Sleet, UNC1069, Dangerous Password, CryptoCore, APT38, BlueNoroff and Stardust Chollima for the North Korean actors; Arctic Wolf specifically attributes the campaign to BlueNoroff, a financially motivated Lazarus subgroup tied to the DPRK Reconnaissance General Bureau and active since at least 2014.

Attack chain: operators take over accounts on Telegram, LinkedIn, Slack and Discord (often belonging to earlier victims) and use them to send video-meeting invitations, reportedly via typosquatted meeting links. During the call the attacker plays recordings of people the target knows (including footage covertly recorded of earlier victims while they waited in the meeting) or deepfakes of prominent crypto/Web3 figures. The attacker then claims a technical fault such as broken audio, shows a fake error and sends a link or instructions for a purported fix (an update or installer). Running it installs malware. Arctic Wolf documents lures including shared financial documents, a fake Chrome PDF viewer update, a fabricated media player called 'NovaStream', identity-verification prompts, hardware-wallet access and security-audit workflows, with custom instruction videos that walk victims past browser warnings on macOS and Windows. On macOS the 'Update Now' button starts an AppleScript download; on Windows/Linux users are told to run commands that fetch a payload from a fake Chrome update site, after which a Chrome-branded 'Update Complete' dialog is shown. Arctic Wolf detection coverage references PowerShell, curl-piped and osascript execution patterns.

Impact and capabilities: the malware steals cryptocurrency data and credentials for video platforms (Zoom, Google Meet, Microsoft Teams), and accesses contact lists, social media, communication apps and webcam video. Victims' contacts and accounts are used to find the next targets, and victims were sometimes robbed or their accounts hijacked months to more than a year after the initial compromise. Arctic Wolf counted 677 targeted individuals across 58 countries (84% founders and C-suite; counts reflect targeting, not confirmed compromise), up from 400+ targets in 20 countries in its April report, and flagged 349 videos as likely modified as of 2026-11-06. Deepfake production moved from Adobe Premiere to generative AI tools (Google Generative AI, Grok Imagine, Wondershare AIGC, GPT-4o), attributed by provenance evidence rather than visual inspection. Arctic Wolf reports a coordinated server cluster (98 high-confidence domains, 73 not previously reported) linked by naming conventions, delivery commands and certificate habits; the domain list is in its PDF report and was not available on the public blog page.

Supply-chain risk: the FBI says the actors are now targeting open-source maintainers. It described a maintainer of a popular JavaScript library (tens of millions of weekly downloads, unnamed) lured into a fake Teams meeting, stopped by quick detection, and confirmed multiple other attempts against npm and Rust crate developers. BleepingComputer also notes an Amazon July 2026 report linking several open-source supply chain attacks to BlueNoroff. The Binance theft (about $6B, attributed to TraderTraitor) and the $1.2B seizure/restraint announced alongside it are a separate matter. Source note: no CVEs, hashes, IPs or domains were published in the public sources reviewed; the FBI page could not be fetched (HTTP 403) and Arctic Wolf's full PDF IOC appendix was not reviewed.

MITRE ATT&CK techniques by tactic:

Execution

T1059.001 PowerShell; T1059.002 AppleScript; T1204.004 Malicious Copy and Paste

Initial Access

T1078 Valid Accounts; T1195.001 Compromise Software Dependencies and Development Tools; T1566.003 Spearphishing via Service

Collection

T1125 Video Capture

Lateral Movement

T1534 Internal Spearphishing

Resource Development

T1583.001 Domains; T1586.001 Social Media Accounts

Impact

T1657 Financial Theft

Defense Evasion

T1684.001 Impersonation

Remediation

Immediate actions

  • Join video meetings only through the official app using the meeting ID/invite code, not links received in chat
  • Verify unexpected meeting invitations through an independent channel, even when they come from a known contact's account
  • Do not run scripts, installers or 'update' prompts offered as fixes for audio/video problems
  • Keep webcams off unless explicitly needed
  • Hunt for osascript, PowerShell and curl-piped execution launched shortly after a video-conference session

Workarounds

  • Restrict clipboard/terminal command execution from browser-driven prompts where feasible via endpoint policy
  • Isolate high-value wallet and signing operations to dedicated hardened devices

Longer-term hardening

  • Train crypto/Web3 staff, executives and open-source maintainers on ClickFix and deepfake-meeting social engineering
  • Enforce phishing-resistant MFA and hardware-backed signing for wallets, package registries and source repositories
  • Alert on abnormal access to webcam, browser credential stores and wallet data
  • Apply npm and crates.io publishing protections (trusted publishing, 2FA, review of release rights)

Affected and fixed versions

  • Multiple — Zoom, Microsoft Teams and Google Meet users (meetings impersonated; credentials targeted)
    Vulnerable versions: Social engineering; not version-specific
  • Multiple — Telegram, LinkedIn, Slack and Discord accounts (hijacked and used to send meeting invites)
    Vulnerable versions: Account takeover; not version-specific
  • Apple / Microsoft / Linux — macOS, Windows and Linux endpoints of crypto/Web3 workers and open-source developers
    Vulnerable versions: User-executed payloads; not version-specific

Timeline

  • Arctic Wolf's earlier (April 2026; exact day not stated) report documented BlueNoroff fake Zoom/Teams meetings with ClickFix prompts and pre-recorded/AI-generated participants, 400+ targets across 20 countries.
  • Amazon threat analysts linked several open-source (npm) supply chain attacks to BlueNoroff in a July 2026 report (exact day not stated).
  • Separate Binance theft (about $6B, attributed to TraderTraitor) occurred on or about this date; the DOJ announced related asset actions alongside FaceStealer.
  • As of this date Arctic Wolf had flagged 349 videos as likely modified (deepfake provenance) and counted 677 targets across 58 countries.
  • FBI published its FaceStealer report/alert describing the attack chain (compromised-account invites, recorded footage, fake error, malicious fix) and the pivot to open-source maintainers, around the same week as the DOJ announcement.
  • DOJ (C.D. Cal.) and FBI Los Angeles announced seizure of FaceStealer's cloud-based infrastructure; FBI seized over 900 victim webcam videos under a federal warrant.
  • BleepingComputer reported the disruption and Arctic Wolf's BlueNoroff attribution.

Sources cited

More from APT38

Drafted by the Threadlinqs research pipeline from 9 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