Cybersecurity · November 2026
The email was real. The website was real. The "are you human?" check was not.
"Check the sender" would not have helped. The sender was real.
On Monday, November 16, our managed detection and response (MDR) service, monitored 24/7, caught an attack in progress at one of our clients. It was detected and contained within about six minutes. We are sharing what happened, without identifying anyone involved, because the technique is spreading and because the usual advice would not have stopped it.
The bait was real
At 1:12 in the afternoon, a member of staff received an email newsletter from a trusted partner organization the client hears from regularly. It was not phishing. The sender was genuine, and so was the email.
The newsletter's button linked to a page on that organization's own, legitimate website. What nobody knew yet was that attackers had quietly injected code into the site. The page loaded normally, then showed a "Verify you are human" check, the same kind of box most of us click through several times a day.
Three keys to "prove" you are human
The check asked for three steps: press the Windows key and R together, then Ctrl and V, then Enter. It read like a slightly unusual human check, nothing more. In fact, clicking the box had already copied a hidden command to the clipboard. Those three keys opened the Windows Run box, pasted the command in, and ran it.
This technique is known as ClickFix, or the fake CAPTCHA. Nobody did anything careless here. A link from a known organization, followed by what looks like a routine prompt, is exactly the situation it is built for.
What the command did
The command ran at about 1:21 PM. It used a decades-old networking tool built into Windows, one most people have never heard of, to download the next stage from the attacker's server. Attackers like built-in tools because ordinary security software tends to trust them.
It then installed a hidden scripting engine and a bundled copy of Java inside a folder named to look like part of Microsoft OneDrive, and added a startup entry with an innocent-sounding name so it would come back after a restart. By about 1:24 PM, roughly three minutes later, the malware was looking up the network's administrator groups. That is the standard first step before an attacker tries to take over a business network.
Six minutes, start to finish
Our MDR service flagged the activity within seconds. By about 1:27 PM, roughly six minutes after the command ran, the computer was isolated from the network. Isolated means the attacker could no longer reach anything through it.
The rest of the afternoon went to confirming the scope and closing it out:
- Passwords were reset and the account's access was restricted
- The server's security logs were pulled and reviewed. They showed no activity beyond that single look-up, and no evidence of any data accessed, other systems affected, accounts created or changes made
- Email logs confirmed nobody else at the client received that newsletter
- The computer is being wiped and rebuilt rather than "cleaned"
- We told the website's owner, who confirmed they are aware and working on it
We also confirmed the source ourselves. Following the same newsletter link, we were shown the same fake human check. The infected page was on the real website, not in a fake email.
Why the usual advice would not have helped
"Check the sender" and "don't click links you weren't expecting" are good habits, and neither would have stopped this. The email was real, and so was the website. What does help is one rule that holds no matter how trustworthy the source looks: no real website will ever ask you to press Windows + R, open PowerShell or Terminal, or paste anything to prove you are human. On a Mac, the same scam asks you to paste into Terminal. If you see it, close the page.
If you have already done it, disconnect from the network straight away by unplugging the cable or turning off Wi-Fi, then call your IT provider. Minutes matter.
For the technical reader
The chain, in order: the Run dialog, then a built-in Windows networking tool downloading a payload, then a scripting engine and bundled Java dropped into a fake OneDrive folder, then a startup entry for persistence, then a lookup of the network's privileged groups for reconnaissance. We have left out the attacker's domain and file names, because they could identify the compromised site.
Prevention sometimes fails. Response limits the damage.
No single control would have guaranteed a different outcome, and it is worth being wary of anyone who claims otherwise. Security awareness training, limiting administrator rights, web and DNS filtering, and good backups all lower the odds and the impact. None of them is a guarantee.
What kept this to one read-only look-up was speed: a team watching around the clock that can isolate a computer within minutes of the first sign of trouble. That is the difference between an incident and a breach. If you are not sure how quickly your organization would know something like this was happening, or who could cut a computer off at 1:27 on a Monday afternoon, that is worth finding out. It is the core of how we approach layered security, and a free assessment is an easy place to start.
The one rule that helps
No real website will ever ask you to press Windows + R, open PowerShell or Terminal, or paste anything to prove you are human. If you see it, close the page.
Get started
Questions about your own setup?
A free assessment turns general guidance into specific next steps for your organization. No pressure, no obligation.