Fake AI Advertising Portals Steal Account Credentials and MFA Codes
Social Engineering & Fraud · 6 sources · Updated 1mo ago · First seen Oct 7, 2026

A phishing campaign is targeting online advertising account managers through fake AI-assisted advertising portals impersonating ChatGPT, Gemini, Claude, Perplexity and other services. The sites promise advertising assistance and display browser-in-the-browser login windows when visitors attempt to connect their accounts. Human operators manage the phishing flow in real time through Telegram, requesting passwords, multi-factor authentication (MFA) codes and additional approval steps. Island researchers reported hundreds of victim submissions to a Telegram control channel, but the number of successfully compromised accounts remains unknown.
Researchers linked the campaign to a broader operation using advertising, recruitment and refund lures through shared infrastructure and source code exposed in public GitHub repositories. Reported components include Next.js and Socket.IO; CyberVeille also reported 73 archived scans across 25 domains associated with one backend, although those counts and its indicators are not independently validated. Stolen advertising accounts could support fraudulent spending, account resale, exposure of client billing information and unauthorized administrator access. Advertising teams should verify portal origins before connecting accounts, adopt phishing-resistant MFA where supported, and review unexpected authentication prompts, administrator changes and advertising spend.


