Human-Operated Phishing Platform Impersonates AI Ad Tools to Steal Credentials and MFA Codes

What Happened

Researchers disclosed a phishing platform described as "human-operated" that impersonates advertising-related products from major AI chatbot brands, including Google Gemini, Anthropic Claude, OpenAI ChatGPT, Perplexity, Meta Muse, and Manus. According to the report, the fake portals present themselves as tools offering ad campaign optimization, ad spend audits, and business-account connections — all designed to lure victims into entering credentials.

Why It Matters

Marketing, growth, and social-media teams are increasingly experimenting with AI-branded ad and business tools, which makes this lure highly plausible to a non-technical audience. Because the platform is described as human-operated rather than fully automated, it may be able to adapt in real time to bypass standard phishing defenses, including intercepting MFA codes to defeat basic two-factor protections. Any organization with staff managing ad accounts or business integrations tied to these AI platforms is a potential target.

What Defenders Should Watch For

• Educate marketing/ops staff that none of the named AI vendors currently require a separate "ad portal" login outside their standard authenticated consoles — treat unsolicited ad-optimization or spend-audit tools referencing these brands with suspicion.
• Monitor for lookalike domains and newly registered domains referencing "ChatGPT ads," "Gemini ads," "Claude business," or similar branded-but-unofficial terms.
• Hunt for anomalous OAuth consent grants or business-account connections to third-party apps claiming AI-vendor affiliation.
• Favor phishing-resistant MFA (FIDO2/WebAuthn) over OTP-based codes where possible, since this campaign reportedly targets MFA codes directly.
• Review email and web proxy logs for clicks on ad-related phishing links tied to AI brand impersonation, and flag credential-entry pages that mimic AI vendor branding.

Developing Intel

This item is based on a single published report and details may evolve as more information becomes available; no CVE or specific detection signature is associated with this campaign at this time. For the original reporting, see The Hacker News.