Fake ChatGPT Ads Use ClickFix to Deploy Remote Access Trojans
Written by AI from the linked story. Not reviewed by an editor. How this works
Threat actors are weaponizing sponsored Google ads and custom ChatGPT variants to trick users into visiting malicious sites that deploy remote access trojan (RAT) malware via ClickFix social engineering techniques. The attack exploits user trust in the legitimate ChatGPT.com domain, making deception significantly harder to detect for the average user. ClickFix attacks typically instruct victims to manually run malicious commands under the guise of fixing a problem, bypassing many traditional security controls. This campaign highlights how AI brand recognition is now being actively weaponized to lower user suspicion and increase infection rates. Organizations without strong end-user training and ad-blocking controls are particularly exposed.
Tactical Insight
Immediate actions
- Block or restrict sponsored/paid search ad results at the DNS or proxy level to reduce exposure to malicious lookalike sites.
- Deploy endpoint detection and response (EDR) tooling configured to flag and block ClickFix-style clipboard and command-execution abuse patterns.
Security awareness & training
- Train employees to verify URLs critically before interacting with AI tools, especially those reached via sponsored search results.
- Conduct phishing simulation exercises that include ClickFix-style social engineering scenarios to build user recognition of the tactic.
- Publish internal guidance reminding staff that legitimate AI platforms never ask users to run terminal or PowerShell commands.
Detection & long-term controls
- Enable DNS filtering and web proxy policies that flag newly registered or lookalike domains impersonating known AI platforms.
- Monitor endpoint telemetry for suspicious execution chains originating from browser processes, particularly clipboard-pasted command execution.
- Implement application allowlisting to prevent unauthorized executables and RAT payloads from running on managed endpoints.