News
Three Minutes From a Breach
| Organization | A small organization with a shared front-desk computer |
| Threat | A "ClickFix" attack delivered through a sponsored search ad |
| Time from click to attack | About three minutes |
| Outcome | Blocked automatically. The malicious software never ran. |
What happened
It started with something ordinary. An employee at the front desk wanted to open ChatGPT, so they searched for it on Google and clicked the first result, just as millions of people do every day. That first result was a paid advertisement placed by criminals.
The ad didn't lead to an obviously fake website. It opened a custom chatbot hosted on the real ChatGPT site, so the web address looked completely legitimate. Within a minute and a half, the chatbot sent them to what looked like a routine security check: a "verify you're human" page, the kind people click through dozens of times a week.
This one was different. It asked the employee to press a few keys to finish the check. What they couldn't see was that the page had secretly copied a hidden command to the clipboard. Following the instructions meant pasting and running that command themselves. It was designed to download malicious software onto the computer, the kind criminals use to steal saved passwords, email access and other accounts.
The command never got that far. The organization's endpoint protection, CrowdStrike Falcon managed by MCS, recognized the attack the moment it ran and stopped it.
The attack, minute by minute
| Minute | What happened |
|---|---|
| 0:00 | An employee searches Google for "ChatGPT" and clicks the top result, a sponsored ad for "ChatGPT 5.6 Plus." |
| 0:05 | The ad opens a custom chatbot on the real ChatGPT website. Nothing about the address looks wrong. |
| 1:30 | The chatbot sends the employee to a "verify you're human" page. Behind the scenes, the page copies a hidden command to the computer's clipboard. |
| 1:45 | The page tells the employee to press a few keys to complete the check: open a Windows command box, paste, and press Enter. The employee reloads the page and tries again. |
| 3:20 | The pasted command tries to download and run malicious software. CrowdStrike Falcon recognizes the behavior and blocks it instantly. |
Why this attack is so effective
- It starts with a normal search. Criminals buy ads that appear above real search results. Clicking the top result is a habit, not a mistake.
- The web address looked real. Because the chatbot lived on the genuine ChatGPT website, the usual advice to "check the address bar" wouldn't have helped.
- It feels like a routine check. People are used to proving they're human online. Pressing a few keys seems like one more step, not a warning sign.
- The victim runs the attack themselves. Because the employee launched the command, it can slip past protections that only watch for files being downloaded or opened. This technique, called "ClickFix," has become one of the most common ways attackers get into business computers.
How it was stopped
Behavior-based protection caught it instantly. Traditional antivirus looks for known bad files. Here there was no file to find yet, only a command typed by a real user. CrowdStrike Falcon watches what programs do, recognized this paste-and-run pattern as an attack, and blocked it before any malicious software was downloaded. It did the same on every retry.
MCS investigated the full chain. Our team contained the computer, traced the attack step by step from the search ad to the blocked command, confirmed the malicious software never ran, and checked that nothing had been set up to relaunch it. We then blocked the attacker's websites, so the same trap can't work elsewhere.
What could have happened instead: stolen passwords and email access, fraudulent payment requests sent from a trusted mailbox, attackers reaching other systems, and days of cleanup and costly downtime. For many small organizations, that's the start of a serious breach.
The one rule that stops this attack
No legitimate website will ever ask you to press Windows + R, open PowerShell or a command window, or paste something to "verify" you or "fix" a problem. If a page asks you to, close it and tell your IT provider.
How to protect your organization
- Bookmark the tools you use. Open ChatGPT, your bank, and business applications from bookmarks rather than searching for them. That skips the ads entirely.
- Treat sponsored results with caution. Scroll past the ads to the regular search results, or go directly to a site you already trust.
- Use behavior-based endpoint protection. Protection that watches what programs do, not just which files they are, is what stopped this attack.
- Train your staff with real examples. Share this case study. People who have seen the trick once are far less likely to fall for it.
- Limit ads and risky browsing on shared computers. Front-desk and shared PCs are frequent targets. Blocking ads on them removes this whole category of attack.
- Put an AI use policy in place. AI tools are now a lure in their own right. A clear policy tells staff which tools are approved and how to use them safely.
How MCS can help
MCS provides managed cybersecurity for small and mid-sized organizations, including:
- A free cybersecurity assessment that reviews the areas attackers target most and gives you a clear, prioritized report.
- CrowdStrike Falcon endpoint protection with monitoring and response from our team.
- Security awareness training using real-world attacks like this one.
- Policy support, including AI use and information security policies.
Schedule your free cybersecurity assessment
Call 217.698.9900 or visit thinkMCS.com.
Details have been changed or omitted to protect the organization's privacy.