Hackers used malicious OpenAI Custom GPTs to trick people into infecting their own Windows PCs after searching for ChatGPT on Google. Huntress says the attackers bought sponsored search placement, named the GPTs to look legitimate, then sent victims to Google Sites pages showing fake Cloudflare checks that told them to run PowerShell commands. At least 40 users were infected; one chain used a Canon-signed app for DLL sideloading, persistence via a Run key and scheduled task, and a remote-access trojan that called out over DNS-over-HTTPS.
Why it matters: This is a live social-engineering malware campaign abusing trusted AI and Google-hosted services, so ordinary users and enterprise defenders alike are at risk. Users should avoid sponsored ChatGPT links and never run pasted PowerShell commands from a web page; defenders should hunt for the described persistence items, MSI downloads, and suspicious DNS-over-HTTPS traffic.
info@thehackernews.com (The Hacker News)
2026.09.30
95% relevant
This appears to be the same underlying campaign: attackers used OpenAI Custom GPTs as part of ClickFix-style lures to trick users into infecting themselves with remote-access malware. This source adds that the payload is a remote-access trojan (RAT) and reinforces the abuse of trusted ChatGPT workflows as the delivery vector.
Bill Toulas
2026.09.29
95% relevant
This is the same underlying event: attackers using OpenAI Custom GPTs promoted via Google ads to redirect users to fake Cloudflare/ClickFix pages that install malware. This article adds details on the specific malicious GPT names, Google Sites staging page, the switch from Canon-signed to Stardock-signed host apps, persistence via Run key and scheduled task named 'Canon Configuration Reader,' and Huntress' observed scope.
Ionut Arghire
2026.09.29
100% relevant
This article establishes a distinct campaign centered on OpenAI Custom GPT abuse as the lure mechanism, with concrete infection details, infrastructure, and observed victim impact.
← Back to all stories