Skip to navigation

Threat reportPhishingTL-2026-5961

Evolving Email-Borne Social Engineering: QR Phishing, ConsentFix, ClickFix/AI-fix/CrashFix and Deepfake Impersonation

mediumACTIVE

ESET reports that email-borne social engineering is shifting toward session/token theft and user-executed commands: QR code phishing (one in nine detected phishing emails in H1 2026), ConsentFix OAuth authorization-code theft against Microsoft sign-in, ClickFix and its AI-fix and CrashFix variants, phishing-as-a-service kits, and AI-assisted lures and deepfake impersonation. First disclosed .

Key facts

IdentifiersTL-2026-5961
Also known asConsentFix, AuthCodeFix, ClickFix, AI-fix, CrashFix
SeverityMEDIUM
First disclosed
Affected productsMicrosoft Entra ID / Microsoft 365 (first-party apps such as Azure CLI, Azure PowerShell, Visual Studio Code); Microsoft Windows (Run dialog / PowerShell / cmd targeted by ClickFix and CrashFix); Google Chrome Web Store / Chrome browser (CrashFix NexShield extension); Apple macOS Terminal (AI-fix shared-chat lures delivering AMOS/MacSync per public reporting)
MalwareAMOS, KongTuke, ModeloRAT, NexShield
Toolsfinger.exe
MITRE ATT&CK techniquesT1059.001 PowerShell, T1078.004 Cloud Accounts, T1176.001 Browser Extensions, T1204.004 Malicious Copy and Paste, T1480.001 Environmental Keying, T1528 Steal Application Access Token, T1547.001 Registry Run Keys / Startup Folder, T1550.001 Application Access Token, T1566.002 Spearphishing Link, T1568.002 Domain Generation Algorithms, T1573.001 Symmetric Cryptography, T1620 Reflective Code Loading and 2 more in the attack flow
MotivationFINANCIAL
Target sectorsfinance, enterprise, small-and-medium-business, technology, government administration
Target regionsGlobal
Detection rules9 (1 shown free)
Sources14 cited sources

Detection coverage

As of 2026-11-13, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-5961 across Splunk SPL, Microsoft KQL and Sigma, covering 25 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: CrashFix/ClickFix finger.exe LOLBin Copied to %TEMP%\ct.exe — Exploitation

Behaviour-first detection of the CrashFix (KongTuke) user-executed command chain: finger.exe is copied to %TEMP%\ct.exe and its output piped to cmd to retrieve a staged PowerShell script that deploys ModeloRAT. Detects a renamed finger.exe via OriginalFileName, the copy-to-temp command line, and finger.exe spawned by cmd.exe. finger.exe use is almost non-existent in modern enterprises, so noise is low.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1204.004, T1059.001
  • Data sources: sysmon (EventCode 1), index=sysmon sourcetype=XmlWinEventLog:Microsoft-Windows-Sysmon/Operational, DeviceProcessEvents, windows process_creation
  • Severity: high
  • Confidence: high

Splunk SPL

index=sysmon sourcetype="XmlWinEventLog:Microsoft-Windows-Sysmon/Operational" EventCode=1 earliest=-24h
(OriginalFileName="finger.exe" OR Image="*\\finger.exe" OR CommandLine="*finger.exe*" OR CommandLine="*\\ct.exe*")
| eval img=lower(Image), cmd=lower(CommandLine), parent=lower(ParentImage), orig=lower(OriginalFileName)
| eval renamed_finger=if(orig="finger.exe" AND NOT match(img,"\\\\finger\\.exe$"),1,0)
| eval copy_to_temp=if(match(cmd,"copy\\s+\\S*finger\\.exe\\s+\\S*(%temp%|\\\\temp\\\\)\\S*\\.exe"),1,0)
| eval finger_from_cmd=if(match(img,"\\\\finger\\.exe$") AND match(parent,"\\\\cmd\\.exe$"),1,0)
| where renamed_finger=1 OR copy_to_temp=1 OR finger_from_cmd=1
| stats count min(_time) as first_seen max(_time) as last_seen values(CommandLine) as command_lines values(Image) as images values(ParentImage) as parent_images by host, User
| convert ctime(first_seen) ctime(last_seen)

Microsoft KQL

DeviceProcessEvents
| where Timestamp > ago(24h)
| where FileName =~ "finger.exe"
    or (ProcessVersionInfoOriginalFileName =~ "finger.exe" and FileName !~ "finger.exe")
    or (ProcessCommandLine has "finger.exe" and ProcessCommandLine has_any ("ct.exe", "%temp%", "\\temp\\"))
| project Timestamp, DeviceName, AccountName, FileName, ProcessVersionInfoOriginalFileName, ProcessCommandLine, InitiatingProcessFileName, InitiatingProcessCommandLine
| order by Timestamp desc

Sigma

title: CrashFix ClickFix finger.exe LOLBin Copy and Execution
id: 5b1d7c2e-3a64-5f0e-9c11-6a2f4d8e7b01
status: experimental
description: Detects the CrashFix/KongTuke ClickFix command chain where the Windows finger.exe LOLBin is copied to %TEMP%\ct.exe or executed from cmd.exe and its output is piped to cmd to fetch a PowerShell stage.
references:
    - https://www.welivesecurity.com/en/business-security/devil-email-wearing-new-mask/
    - https://blog.nviso.eu/2026/01/29/consentfix-a-k-a-authcodefix-detecting-oauth2-authorization-code-phishing/
    - https://www.huntress.com/blog/malicious-browser-extention-crashfix-kongtuke
    - https://pushsecurity.com/blog/consentfix
author: Threadlinqs Team
date: 2026-11-13
tags:
    - attack.execution
    - attack.t1204.004
    - attack.t1059.001
    - attack.t1105
logsource:
    category: process_creation
    product: windows
detection:
    selection_renamed:
        OriginalFileName: 'finger.exe'
    filter_real_name:
        Image|endswith: '\finger.exe'
    selection_copy:
        CommandLine|contains|all:
            - 'copy'
            - 'finger.exe'
            - 'ct.exe'
    selection_cmd_child:
        Image|endswith: '\finger.exe'
        ParentImage|endswith: '\cmd.exe'
    condition: (selection_renamed and not filter_real_name) or selection_copy or selection_cmd_child
falsepositives:
    - Administrators legitimately querying finger services (very rare)
level: high

Known false positives:

  • Legacy administrator scripts that intentionally query finger daemons
  • Security testing or red-team emulation of ClickFix

Tuning notes: Requires Sysmon EventCode 1 with OriginalFileName populated. Allowlist known admin hosts/accounts if finger is used legitimately. Pair with outbound TCP/79 blocking.

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)
  • ConsentFix OAuth Authorization-Code Theft — Interactive/Non-Interactive Session Mismatch — Exploitation · unlock with TL Creds
  • ConsentFix Stolen-Token Use — First-Time Abused First-Party App by User — Post-Exploitation · unlock with TL Creds
  • ModeloRAT 300-Second Python Beacon Pattern — C2 · unlock with TL Creds
  • ClickFix Pasted Command in Run Dialog (RunMRU) — Exploitation · unlock with TL Creds
  • CrashFix Staged PowerShell Host Fingerprinting and AMSI Bypass — Post-Exploitation · unlock with TL Creds
  • ConsentFix Delivery, temp.sh Staging and stream*.cyou Malvertising Domains — C2/Delivery · unlock with TL Creds
  • ModeloRAT Run-Key Persistence and NexShield Extension Artifacts — Post-Exploitation · unlock with TL Creds
  • CrashFix ModeloRAT C2 IPs, finger TCP/79 and NexShield C2 Domains — C2 · unlock with TL Creds

Indicators of compromise

As of 2026-11-13, Threadlinqs Intelligence tracks 25 indicators of compromise for TL-2026-5961.

Sample indicators (3 of 25)

  • IP address: 170.168.103[.]208
  • Domain: nexsnield[.]com
  • URL: hxxps://www[.]dropbox[.]com/scl/fi/6gscgf35byvflw4y6x4i0/b1.zip?rlkey=bk2hvxvw53ggzhbjiftppej50&st=yyxnfu71&dl=1

Defanged, TLP:CLEAR. Full set of 25 indicators: unlock with TL Creds.

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

Technical analysis

This record documents a trend overview published by ESET (WeLiveSecurity, 2026-09-28, author Tomas Foltyn) describing how email and web-borne social engineering has moved away from password harvesting toward techniques that either steal already-authenticated sessions/tokens or make the victim execute the attacker's command themselves. The ESET article itself names no CVEs, threat actors, malware families or IOCs; the technical detail and indicators below are drawn from the primary research that the described techniques correspond to (Push Security, NVISO, Huntress, Cloud Security Alliance) and are labelled as such.

QR code phishing: per ESET telemetry, QR codes accounted for one in nine phishing emails detected in H1 2026, and Microsoft ranks QR codes as the fastest-growing email-based attack vector. The destination URL is hidden inside the image and is opened on a personal phone, bypassing desktop email-gateway URL rewriting and endpoint controls.

ConsentFix (a.k.a. AuthCodeFix): the victim lands on a compromised but legitimate website showing a fake CAPTCHA/Cloudflare-Turnstile-style gate that filters targets by email address. The victim is sent through a genuine Microsoft sign-in (login.microsoftonline.com) for a pre-consented first-party application, most commonly Azure CLI (client_id 04b07795-8ddb-461a-bbee-02f9e1bf7b46), with prompt=select_account and the victim's login_hint. Because an active Microsoft session exists, no password or MFA prompt appears; the browser is redirected to http://localhost:<port>/?code=<authorization_code>&session_state=<id>, which fails to load because nothing listens there. The victim is told to paste that URL back into the phishing page; the attacker redeems the authorization code at the Entra token endpoint for access and refresh tokens (scope https://graph.microsoft.com/.default) and verifies access with GET /v1.0/me. Because the code is issued after authentication has completed, passwords and even phishing-resistant MFA do not stop it. NVISO lists other first-party apps with Conditional Access bypasses (Azure PowerShell, Visual Studio, VS Code, SharePoint Online Management Shell, Teams and others) and a KQL detection joining SigninLogs and AADNonInteractiveUserSignInLogs on user, AppId and SessionId. The Cloud Security Alliance describes a ConsentFix v3 variant active as of May 2026 with an automated backend pipeline.

ClickFix and variants: ClickFix tricks users into pasting an attacker-supplied command into their own terminal (Run dialog / PowerShell / macOS Terminal); ESET notes detections are rising. AI-fix places fake troubleshooting instructions on legitimate domains belonging to Anthropic, OpenAI and Microsoft (publicly shareable AI chat pages), lending the lure the trust of a legitimate domain; public reporting ties such shared-chat lures to macOS infostealers (AMOS, MacSync). CrashFix is a fake ad-blocker browser extension that directs targets to the official Chrome Web Store and waits an hour after installation before showing its first bogus alert. Huntress attributes the CrashFix campaign to KongTuke: the NexShield extension (a uBlock Origin Lite clone, extension ID cpcdkmjddocikjdkbbeiaafnpdbdafmi) exhausts browser resources to force a crash, then on restart shows a fake 'browser stopped abnormally' scan prompt instructing the user to press Win+R and paste a clipboard-loaded command. The command copies finger.exe to %TEMP%\ct.exe and pipes its output to cmd, fetching a staged PowerShell script that fingerprints the host (50+ analysis tools/VM checks), profiles it as domain-joined or WORKGROUP, and for domain-joined hosts deploys the Python-based ModeloRAT (RC4-encrypted C2, Run-key persistence 'MonitoringService', 300-second beacon). Non-domain hosts follow a DGA/AMSI-bypass branch that was still in testing at time of analysis.

Deepfake impersonation and AI misuse: ESET notes AI tools clean up lure language, tailor messages per recipient and build rapport over time, and cites a case in which a finance employee joined a call populated by deepfake versions of senior colleagues and made wire transfers exceeding US$25 million (CNN reporting, 2024). ESET also notes phishing-as-a-service kits are sold ready-made but names none.

Defensive takeaway: controls must cover token/session theft (monitor interactive-vs-non-interactive sign-in mismatches for first-party apps, Azure CLI sign-ins by non-admins, AADGraphActivityLogs), user-initiated command execution (Run-dialog/clipboard-to-shell chains, finger.exe use), and out-of-band verification for payment requests. ESET MDR Q1 2026 also found employees often delete phishing they recognise because no reporting channel exists.

MITRE ATT&CK techniques by tactic:

Execution

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

Defense Evasion

T1078.004 Cloud Accounts; T1480.001 Environmental Keying; T1620 Reflective Code Loading; T1684.001 Impersonation

Persistence

T1176.001 Browser Extensions; T1547.001 Registry Run Keys / Startup Folder

Credential Access

T1528 Steal Application Access Token

Lateral Movement

T1550.001 Application Access Token

Initial Access

T1566.002 Spearphishing Link

Command and Control

T1568.002 Domain Generation Algorithms; T1573.001 Symmetric Cryptography

defense-impairment

T1685 Disable or Modify Tools

Remediation

Immediate actions

  • Hunt Entra ID SigninLogs/AADNonInteractiveUserSignInLogs for Azure CLI (04b07795-8ddb-461a-bbee-02f9e1bf7b46) and other first-party app IDs where a non-interactive sign-in with the same SessionId follows an interactive one by 10-600 seconds from a different IP or country
  • Investigate Azure CLI sign-ins by users who are not administrators or developers, and sign-ins to the 'Windows Azure Active Directory' resource
  • Revoke refresh tokens/sessions for any user who pasted a localhost?code= URL into a web page
  • Search endpoints for finger.exe copied to %TEMP%\ct.exe, Run-dialog-initiated cmd/PowerShell chains, and the HKCU Run value MonitoringService
  • Remove the NexShield extension (ID cpcdkmjddocikjdkbbeiaafnpdbdafmi) and block the listed C2 domains/IPs

Workarounds

  • Block outbound finger.exe/TCP 79 traffic and restrict the Run dialog or clipboard-to-shell workflows via policy where feasible
  • Treat QR codes in email as untrusted and open them only on managed devices with web filtering

Longer-term hardening

  • Enable AADGraphActivityLogs and alert on Entra ID enumeration via legacy scopes
  • Restrict or monitor consent to first-party Microsoft apps and apply Conditional Access with token protection/session controls where available
  • Restrict browser extension installation via allowlist policy
  • Provide an easy phishing-reporting channel and train staff on ClickFix, QR phishing and deepfake-call verification
  • Require out-of-band callback verification for payment and credential requests, including on video calls

Affected and fixed versions

  • Microsoft — Entra ID / Microsoft 365 (first-party apps such as Azure CLI, Azure PowerShell, Visual Studio Code)
    Vulnerable versions: All tenants with default first-party app trust; no patch
  • Microsoft — Windows (Run dialog / PowerShell / cmd targeted by ClickFix and CrashFix)
    Vulnerable versions: Behavioral abuse, not a software flaw
  • Google — Chrome Web Store / Chrome browser (CrashFix NexShield extension)
    Vulnerable versions: Extension 2025.1116.1842 (since removed)
  • Apple — macOS Terminal (AI-fix shared-chat lures delivering AMOS/MacSync per public reporting)
    Vulnerable versions: Behavioral abuse, not a software flaw

Timeline

  • CNN reports a Hong Kong finance employee wired over US$25 million after a video call populated by deepfake versions of senior colleagues; ESET later cites it as an example of deepfake impersonation.
  • uBlock Origin Lite release 2025.1116.1841 is published; the NexShield extension used in CrashFix is cloned from it (version 2025.1116.1842).
  • Per Huntress, KongTuke begins using the finger.exe LOLBin in ClickFix-style commands (December 2025; exact day not stated, 1st used as placeholder for the month).
  • Push Security publishes the first public analysis of ConsentFix, an OAuth authorization-code theft technique abusing the Azure CLI first-party app.
  • NexShield extension (CrashFix) is live on the Chrome Web Store, promoted via a Google ad for ad blockers.
  • Huntress publishes 'Dissecting CrashFix: KongTuke's New Toy', documenting the delayed-crash lure and ModeloRAT delivery.
  • NVISO publishes detection guidance for ConsentFix/AuthCodeFix, listing affected first-party app IDs and a KQL query (updated 2026-06-10).
  • Cloud Security Alliance research note describes ConsentFix v3 with an automated backend pipeline, active as of May 2026.
  • ESET publishes 'The devil is still in the email, but wears a new mask', reporting QR codes in one in nine detected phishing emails in H1 2026 and the rise of ConsentFix, ClickFix, AI-fix and CrashFix.

Sources cited

Previous and next in phishing

Drafted by the Threadlinqs research pipeline from 14 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