850+ fake ChatGPT ads on Google trick Windows users into installing malware
Malvertising campaigns are moving over to chatbots.
Image by Cybernews
Follow us
- Google ads for fake ChatGPT pages led Windows users toward malware traps.
- Researchers linked the campaign to around 850 paid-ad landings and 26 lookalike ChatGPT destinations.
- The scam used fake verification pages that told people to press Win+R and paste commands.
- No legitimate CAPTCHA should ask you to run Windows commands or PowerShell to prove you are human.
Key Takeaways by nexos.ai, reviewed by Cybernews staff.
A routine Google search for ChatGPT is being turned into a malware trap targeting Windows users, researchers have warned.
ChatGPT allows people to create Custom GPTs and share conversations and content. However, cybercriminals are abusing the feature to scam netizens.
Researchers at Island say that attackers are creating their own content inside ChatGPT – via CustomGPT – and then buying Google ads to direct people to it.
Once users start to engage with the attacker-created GPT, instead of receiving a relevant answer, they’re repeatedly shown the same message claiming the service was experiencing "high traffic” and directing them to a supposed “backup domain.”
Island notes that the identical response appeared across conversations from 2 separate lookalike Custom GPTs over 3 days.
The sprawling malvertising campaign involved around 850 paid-ad landings, 26 lookalike ChatGPT destinations, and 71 Google Ads campaign IDs over a 3-month period, ending in August.
ClickFix trap
Anyone following the link could be bounced to a counterfeit ChatGPT and Cloudflare verification page where users are instructed to press Win+R, paste clipboard content, and press Enter – supposedly to prove they are human.
Island warns that this is a “ClickFix” trap – a social engineering technique that tricks people into executing malicious commands on their own devices.
The instructions trick Windows into running a hidden command that downloads disguised malware onto the computer.
Once there, the malware can infect the system, survive restarts, and secretly communicate with attackers through a Telegram bot.
The malware also uses a well-known tool called NetSupport RAT, which lets an attacker remotely control the infected computer as if they were sitting in front of it.
Researchers also saw the attackers rotate domains and use other Windows malware-delivery routes, suggesting the scam was not dependent on a single disposable website.
Island says one of its AI tools first noticed the suspicious activity. Human researchers then confirmed what it was and how it worked.
While the researchers were able to track the ads, the GPT lures, the malware delivery mechanisms, and the scale of the campaign, its data does not show how many users were affected.
How to prevent being scammed by malvertising
Malvertising has become a common attack vector, with criminals exploiting the accountability vacuum.
A single ad click triggers a chain of providers, domains, and servers – and each one has an excuse.
Ad networks blame cloaked content, trackers call themselves "just analytics," publishers point to third parties, and hosts claim ignorance.
Malvertising is now so common that some want cyber insurers to mandate adblockers.
For Windows users, there’s one huge red flag in the campaign Island has spotted: no CAPTCHA or routine website verification should require you to open Windows Run and paste a command.
If a website tells you to press Win + R, paste something, or run PowerShell to prove you’re human, don’t do it.
