deniz.in

Markets

Weather

Loading weather

· via dev.to (home feed)

ClickFix attacks use fake CAPTCHAs, clipboard commands and PNG payloads to reach domain controllers

Fake human-verification pages now place malicious commands on the clipboard, hide programs inside PNG pixels, and have carried intrusions from single PCs to Windows domain controllers, according to Huntress.

ClickFix attacks use fake CAPTCHAs, clipboard commands and PNG payloads to reach domain controllers

Fake "verify you are human" pages that tell visitors to paste a command into Windows have moved well beyond a consumer nuisance. According to a Huntress report published on 10 November 2026, the social-engineering technique known as ClickFix has grown more polished and now includes payloads hidden inside PNG images and intrusions that travelled from a single infected PC to domain controllers. The findings were summarised in a write-up from security firm Coinetic AI Security republished on dev.to; the vendor also sells a product it says terminates the pasted commands as they launch.

How the attack starts

Victims land on the malicious pages through lookalike domains, manipulated search advertising, email links or hacked legitimate websites — the post cites one compromised dental practice site. The page shows a familiar Cloudflare-style human verification check, or an error claiming something must be fixed to view a document. Interacting with it silently places a command in the clipboard.

The page then instructs the user to press Windows+R, or open a terminal via Windows+X, paste with Ctrl+V and hit Enter. The command is deliberately long and padded so the visible portion looks harmless while the dangerous part sits out of view. Once executed, it uses built-in Windows tooling — PowerShell, mshta, curl, certutil or msiexec — to fetch the next stage. Newer versions add a fake verification progress bar to occupy the victim while the download runs.

Because the victim personally runs the command from a native Windows interface, much security software logs it as routine administrative activity rather than an attack.

Payloads hidden in pictures

One observed chain concealed complete programs inside the pixel data of PNG images. A PowerShell stager rebuilt those programs on disk and executed them. From a file scanner's point of view, the downloaded material consists of nothing but pictures.

From one PC to the domain controller

Huntress documented intrusions that escalated from the first compromised machine to domain controllers over the course of 2026 using standard Windows remote administration tools: PsExec in February, SMBExec over administrative shares reaching 11 hosts in May, and in September WinRM against the primary domain controller, followed by a reverse SSH tunnel out of the network.

Malware families seen in these campaigns include Matanbuchus 3.0, AstarionRAT and DuskDesk, described in the post as a Node.js remote access tool disguised as management software. Persistence typically relies on scheduled tasks and Run keys, and the malware steals browser passwords and session cookies.

Recommended defences

The post's guidance splits into habits and hardening. For individuals: never paste anything a website provides into the Run box, a terminal or PowerShell — no legitimate site asks for this — run as a standard rather than administrator account, and keep systems updated.

For organisations, the recommendations include application control through WDAC or AppLocker to block execution from user-writable folders such as Downloads, AppData and Temp; removing local admin rights and deploying LAPS; and enabling PowerShell script block logging, which records commands as event 4104. Detection, the post argues, should focus on behaviour rather than file names: browsers, Office applications or PDF readers spawning PowerShell, cmd, mshta or Windows Terminal; encoded PowerShell invocations; new scheduled tasks (event 4698); new services (7045); and command-line auditing via event 4688.

To keep one infected endpoint away from the domain controller, the post advises restricting WinRM on ports 5985 and 5986, SMB and remote service management to designated admin machines via the Windows Firewall — while noting that domain controllers still need SMB from ordinary PCs for logon scripts and Group Policy. It also suggests placing administrators in the Protected Users group, keeping separate admin accounts, and blocking outbound TCP 22 from servers, since the September case exfiltrated through a reverse SSH tunnel.

Why it matters

ClickFix inverts the usual malware model: rather than exploiting a software flaw, it persuades the user to run the attacker's command personally, which blinds defences that trust user-initiated actions. PNG steganography means download scanners see images rather than executables. And the documented path from one clipboard paste to domain controller compromise shows how far a single social-engineering click can travel in a flat network. The countermeasures are cheap and well understood — refusing to paste commands, dropping local admin rights, and constraining remote administration protocols — which is exactly why the attack leans on users and lax defaults rather than novel technical exploits.

  • #social-engineering
  • #malware
  • #windows
  • #phishing
  • #security

Related posts