Malicious Custom ChatGPTs Deliver RAT Malware via ClickFix Attacks

Threat actors are abusing custom versions of ChatGPT, promoted through sponsored Google search results, to direct users to malicious sites carrying out ClickFix attacks. The malicious model, named "Plus 5.6," tells users to visit an alleged backup site hosted on Google Sites, where a fake Cloudflare verification page instructs victims to run a PowerShell command. Because the instructions originate on the legitimate ChatGPT.com domain, researchers say victims are more likely to comply.

If executed, the PowerShell command installs a malicious MSI file that loads a remote access trojan alongside a legitimate, signed application via DLL side-loading. The RAT provides remote desktop access, audio and camera capture, file searches, host reconnaissance, and the ability to run additional payloads. It achieves persistence by creating a Windows Registry Run key and a scheduled task, both disguised as "Canon Configuration Reader."

Huntress, which discovered the campaign, investigates at least 40 incidents connected to the Google Sites page, though only two confirm use of the custom GPT variant. OpenAI removes the first malicious GPT by September 25, but a second linked GPT surfaces two days later and remains active at the time of reporting. More recent attacks switch from a Canon-signed host application to one signed by Stardock, showing the operators continue evolving their techniques. OpenAI plans to retire custom GPTs entirely on December 11.

Read More at the original source →