Criminals have been paying for Google adverts on searches for “ChatGPT” that lead, through a chain of trusted-looking pages, to a request to paste a command into Windows. Anyone who does it hands the criminals remote control of their computer.
What’s happened
Security company Huntress published details of the campaign on 28 September. The adverts appeared when people searched for “chatgpt” and led to a fake custom chatbot called “Plus 5.6”, hosted on ChatGPT’s own website and named to look like a new ChatGPT model.
The chatbot sent people to a page on Google Sites dressed up as a Cloudflare “are you human?” check. It told them to copy a command and paste it into Windows PowerShell to prove they were human. Security researchers call this trick ClickFix.
The command installed malware that gave the criminals remote control of the PC, including the screen, the camera and microphone, saved browser data and the ability to install more malware. Huntress says it responded to at least 40 incidents linked to the campaign. OpenAI removed the first fake chatbot by 25 September, but another appeared on 27 September.
What it means for your business
Every step looks legitimate: a Google search, the real ChatGPT website and a familiar security check. Staff who use AI tools to get work done are exactly the people it’s aimed at, and antivirus doesn’t always stop a command that the user runs themselves.
Once this kind of malware is on a work PC, criminals can see everything on the screen, including email, online banking and client files, and use it as a way into the rest of your network.
What to do now
- Share one simple rule with your team: no genuine website will ever ask you to paste a command into your computer. If a page asks you to, close it and tell whoever looks after your IT.
- Bookmark the AI tools your business has approved and open them from the bookmark instead of searching for them. The sponsored results at the top of a search are adverts, and criminals can buy them too.
- Make sure every PC has security software that someone actively monitors, so unusual activity is spotted and dealt with quickly.
- If someone has already run a command like this, disconnect the PC from the network and call your IT support straight away.
Related articles
- Does your business need an AI acceptable use policy?
- ChatGPT, Copilot, Claude or Gemini? How business AI plans handle your data
Sources
- Huntress: Attackers abuse ChatGPT custom GPTs to deliver malware via ClickFix (28 September 2026)
- BleepingComputer: Custom ChatGPTs push ClickFix attacks to deploy malware (29 September 2026)
Details are correct at the time of writing. Check the sources above for any later changes.
Want us to check this for your business? Call 01787 209 003, see our cyber security service, or book a free IT consultation.