Connect with us

NEWS

ClickFix Turns the Human Check Into a Malware Installer

ClickFix fake CAPTCHA pages exploit the habit of proving you are human, then have you paste a command that installs malware on your own PC.

Published

on

A fake CAPTCHA that asks you to press Windows+R, paste, and hit Enter now puts malware on the PC you just verified as human. Microsoft Defender saw that pasted command run on more than 1.1 million unique devices between February and early May 2026, about eight times the weekly count at the start of that stretch.

Chris Fistonich, a researcher at Malwarebytes, says ClickFix is the most popular way to get malware onto a computer. The lure fits Christmas shopping, when people look up small shops they have heard about and hit a human check they have been trained, for years, to complete without reading.

A Test Built to Stop Bots Now Recruits Humans

In 2000, Yahoo asked a Carnegie Mellon team that included Luis von Ahn for a way to stop programs from opening piles of fake email accounts, wrecking search results, and jumping ticket queues. The group built a test that humans could pass but computers could not, using warped text a person could read and a machine of that era could not. Yahoo put it on sign-up pages in 2001. The name, Completely Automated Public Turing test to tell Computers and Humans Apart, told you the job.

WHAT CAPTCHA WAS BUILT TO DO

  • The target: Bots that opened fake accounts, stuffed search, or grabbed tickets.
  • The method: A puzzle a person could solve and a program could grade.
  • The spread: Distorted text, then image clicks, then quiet behavior checks inside the browser.

That ritual worked because people learned to obey it. Click a box. Pick the buses. Type the grainy letters. The page is supposed to stay in the browser. ClickFix keeps the familiar prompt and moves the real work onto the operating system, where a pasted line can launch PowerShell, cmd.exe, or another tool Windows already trusts.

The Three Keys That Install the Malware

The page may say “Verify you are human by completing the action below” and ask you to tick a box. That click is often the moment JavaScript copies a command onto the clipboard with navigator.clipboard.writeText, which you never see. Then the instructions leave the browser.

One lure Fistonich’s team captured reads: “To better prove you are not a robot, please: Press & hold the Windows Key + R. In the verification windows, press Ctrl + V. Press Enter on your keyboard to finish.” Other pages say something went wrong while displaying the site and offer the same three keys as the fix.

THE SEQUENCE THE PAGE WANTS

  • Windows+R: Opens the Run box, a small trusted window most people rarely use and do not treat as a command line.
  • Ctrl+V: Pastes whatever the page already placed on the clipboard, which is not the short “verification” text on screen.
  • Enter: Runs that line with the signed-in user’s rights, often as a shortened PowerShell download that pulls the rest from the network.

Microsoft Threat Intelligence says operators bank on the fact that most targets do not know what Run is for, unlike administrators who live in it. The box accepts a practical maximum of 259 characters, so the first command is only a stub. Seeing the fake check, or even having the command land on the clipboard, does not infect the machine. Infection starts when you paste and press Enter.

People are not prepared for this. They are used to solving Captchas every day. They’re used to being able to trust their security software. So it really has opened a new security loophole that’s being exploited extensively.

Chris Fistonich, researcher, Malwarebytes

Fistonich calls it one of the most sophisticated malware campaigns in use, not because the code is exotic, but because it rides a habit security products were not built to argue with. The user is the installer.

Microsoft Counted More Than 1.1 Million Devices in Spring

The 2026 Microsoft Digital Defense Report, covering July 2025 through June 2026, says ClickFix-style pasted commands ran on more than 1.1 million unique devices between February and early May 2026. Weekly volume in Defender telemetry rose from 38,000 devices in the first week of February to 310,000 in the first week of May, about eight times over that window. Microsoft says the follow-on is almost uniformly malware.

CLICKFIX VOLUME IN MICROSOFT DEFENDER TELEMETRY

Measure Figure Period
Devices per week running a pasted command 38,000 to 310,000 First week of February to first week of May 2026
Growth across that spring window About eight times February to early May 2026
Rebound after a late-2025 dip Roughly 23 times December 2025 to May 2026
ClickFix-style share of initial access 47 percent Prior Digital Defense Report year
Malicious copy-and-paste share of observed intrusions 12.8 percent July 2025 through June 2026
Those intrusions that then loaded malware 96.3 percent July 2025 through June 2026

The 47 percent figure is last year’s share of initial access in Defender Experts notifications, when ClickFix outranked phishing. This year’s 12.8 percent is a different slice, copy-and-paste as a share of observed intrusions, in a year when other paths rose too. Absolute volume still exploded. Microsoft calls it one of the most conversion-efficient ways in that it sees, and it ties the spring spike to kits that anyone can rent.

A $200 Kit Turns Any Site Into a Captcha Trap

Microsoft first saw the method between March and June 2024, in email from a cluster it tracks as Storm-1607, with HTML attachments that tried to install the DarkGate loader. By August 2025, campaigns were hitting thousands of business and home devices every day. On hacker forums since late 2024, ClickFix builders sold for US$200 to US$1,500 a month, often bundled with LNK, JavaScript, and SVG generators and Cloudflare-styled landing pages. Sellers also offered one-time source, pages, or command lines for US$200 to US$500, and advertised antivirus and SmartScreen bypass plus persistence.

Arrival is ordinary. A phishing mail, a poisoned advert, or a real site that has been broken can all drop the same card. Microsoft has seen fake Word errors, fake Discord servers, and brand impersonations including Booking.com. In May 2025 a Lampion banking-stealer run used a fake Portuguese tax site after a ZIP that held an HTML file. The payload mix is a catalogue, not a single gang.

WHAT USUALLY LANDS AFTER THE PASTE

  • Infostealers: Lumma Stealer is the payload Microsoft has seen most, taking saved passwords, cookies, and session tokens that can open accounts without a second prompt.
  • Remote tools: Xworm, AsyncRAT, NetSupport, and SectopRAT give hands-on control for later movement.
  • Loaders and rootkits: Latrodectus, MintsLoader, and a modified r77 rootkit stage whatever comes next, often in memory rather than as a visible.exe.
  • Mac stealers: A June 2025 run aimed at macOS copied a different clipboard command and dropped Atomic macOS Stealer, including Poseidon and Odyssey variants.

MITRE now files the pattern as T1204.004, a malicious copy and paste technique created on 18 March 2025, covering Windows, macOS, and Linux. The entry notes that a user who runs the line can slip past mail filters, browser sandboxes, and download blockers built for files. Kimsuky and MuddyWater have used the same trick, which means the rental kit is no longer only a crimeware product.

The Paste Moved Off the Run Box

Once defenders started watching Run, the instructions changed. The social engineering stayed. Only the window moved.

HOW THE LURE SHIFTED

  1. March to June 2024: Microsoft sees HTML mail that copies a command and tells the user to paste it into Terminal or PowerShell.
  2. June 2025: macOS pages still show Windows-style steps on screen while the clipboard holds a curl-and-sudo chain that steals the password, strips quarantine, and launches AMOS.
  3. February 2026: A wide campaign tells people to press Windows+X then I, which opens Windows Terminal directly and skips detections tuned to Run.
  4. Spring 2026: FileFix tells users to paste into the File Explorer address bar instead, a pool of people who never open Run or Terminal.
  5. 3 October 2026: Compromised sites pre-load a script into the browser cache disguised as a PNG, so the short Run line only has to find that cached file and rename it as a VBScript.
  6. 20 October 2026: Chrome adds a ClickFix copy warning, but only for users who turned on Safe Browsing Enhanced Protection.
  7. 12 November 2026: Flare’s emerging threats team finds live fake checks delivering previously undocumented tools, including a RAT it named CrashHandler, malware that talks through the Stripe API, and CrocoRat instructions hidden in DNS answers.

Microsoft Threat Intelligence laid out the cache trick in public, including the hunt advice that a CAPTCHA should not ask anyone to run code.

The same week, SANS researchers described a macOS fake check that told users to paste into Terminal and pull a Go infostealer they called DownTheDRAIN, which hunted wallet addresses and Keychain data. In their test, macOS 26.7.3 dumped the file in Trash before it sent anything out. The paste instruction was still the hole. Huntress has been writing up chains that no longer stop at a stealer, and instead set persistence, look around the network, and lift credentials. TikTok added another path in 2025, with faceless clips of fake activation steps; Microsoft says some of those videos passed about 500,000 views and ended in Vidar or StealC.

Chrome Warns After the Command Hits the Clipboard

Jasika Bawa, group product manager for Chrome, said on 20 October 2026 that ClickFix moves the download into PowerShell or cmd.exe, which means Chrome’s own download protections never see it. For Enhanced Protection users, more than 1 billion of them when Google last shared that figure in 2025, the browser now watches for a suspicious command on copy. The clipboard text is stripped of personal details, sent to Safe Browsing, and if the verdict is dangerous, Chrome shows a warning after a dangerous verdict. Google says the check is tuned to spare people who really are copying developer commands. It does not say the bad text is removed from the clipboard. Standard Protection does not get this step. Edge does not use this Safe Browsing path.

Microsoft’s own write-up still recommends the blunt control: take Run away from staff who never need it, turn on PowerShell script-block logging, and use application control so native binaries cannot be launched from that box. Apple has already put a malware paste warning in Terminal. Windows still offers Run to every user who can find Windows+R. That gap is why the spring numbers look the way they do. A warning in one browser, on one opt-in mode, does not cover the person who followed a cloned Cloudflare card on a hijacked shop and pressed Enter before anyone could argue.

What a Real CAPTCHA Never Asks You to Do

A genuine Google or Cloudflare check stays inside the browser. It never asks you to open Windows Run, Windows Terminal, PowerShell, or Mac Terminal, and it never asks you to paste a command. If the page wants those keys, close it. Security software can block some of the trap sites, which is worth having, but it will not always stop a command you ran yourself.

IF YOU ALREADY PRESSED ENTER

  • Assume theft, not a scare: Infostealers grab browser cookies and saved logins within minutes, including sessions that skip a password prompt.
  • Call the bank first: Fistonich says to contact the bank immediately, then change passwords, starting with email and anything tied to money.
  • Use a clean device: Reset those passwords and sign out other sessions from a machine that did not run the command.
  • Tell work if this is a work PC: A home paste can still expose office mail, cloud files, and single sign-on tokens.

The next page may look like a gift shop, a tax form, a video meeting, or a Cloudflare wait screen. The dangerous step will still be the paste. If a human check asks you to prove it with Windows+R, you have already seen the tell.

Disclaimer: This article is news reporting and analysis of a malware delivery method, and it is for information only. It is not cybersecurity, legal, or financial advice, and it is not a guide to cleaning an infected computer or recovering stolen funds. Anyone who thinks a device has run a pasted command should speak with a qualified IT or security professional, and with their bank if payment or login details may have been exposed. Figures, product names, and campaign details reflect the public sources cited here and can change as operators switch lures and as vendors update their tools.

Harry is the editor and lead writer of CUMBERNAULD MEDIA, which he runs as an independent publication after a decade in journalism spent moving from reporting to editing. His habit is to open the document before the summary of it. A company result is read from the filing rather than the press release, a court or regulatory decision from the judgment itself, a scientific finding from the paper and its methods section rather than the headline claim, and a sporting sanction from the governing body's own ruling. That approach shapes coverage across news, business and technology as much as science, sports and entertainment, and it carries into the lifestyle, travel, auto and gaming pages, where product specifications are checked against the manufacturer's sheet and, where possible, against Harry's own testing. Every number is checked before publication, and where a source's figures are disputed the story says so. Corrections follow a public policy and are marked on the page. Readers anywhere in the world who write in get a reply from him, and the address is support@cumbernauld-media.com.

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending