From Hitting Connect to Getting Fake AI Ads: DNS Deep Dive
Island security researchers¹ uncovered a human-operated phishing platform posing as a suite of AI advertising tools. Its pages impersonate ChatGPT, Claude, Gemini, and Perplexity. Clicking “Connect” opens a fake sign-in window drawn inside the phishing page, where a live operator captures passwords and MFA codes. The prize is the victim’s ad account.
The Island analysis published 137 network IoCs—30 subdomains and 107 domains. We investigated them with the WhoisXML API MCP Server², examining registration records, DNS histories, hosting, and connected artifacts. That left 132 IoCs for analysis after filtering out legitimate and inactive domains.
Original IoCs Analyzed
Subdomains and domains from the reported campaign
Key Findings
Email-connected domains
IP addresses
String-connected domains
Malicious artifacts
These are key highlights from our analysis. The full report contains many more findings and detailed insights.
OUR RESEARCH PROCESS
- Subdomain IoC hosting and threat analysis
- Domain IoC legitimacy check
- Domain IoC WHOIS and DNS history analysis
- Email-connected domain discovery
- IP address resolution and geolocation analysis
- String-connected domain discovery and malicious artifact confirmation
Download a sample of the threat research materials now or contact sales to discuss your intelligence needs for threat detection and response or other cybersecurity use cases.
—
- [1] https://www.island.io/blog/behind-the-connect-button-the-fake-ai-ads-campaign
- [2] https://main.whoisxmlapi.com/ai/mcp-server
Latest Reports
Integrity Technology Group-Backed Actors Combine Automated and Manual Tools in Global Attack
WhoisXML API uncovered 7,100+ additional artifacts linked to the Integrity Tech-enabled global data theft attack. Download the threat research materials now.
Earth Sirrush Espionage Tool Evolution: DNS Facts Unveiled
WhoisXML API uncovered 400+ additional artifacts linked to Earth Sirrush. Download the threat research materials now.
Four hours before PHP2Shell’s Disclosure
A server scouted our honeypots 4 hours before PHP2Shell’s disclosure, then returned with exploit attempts. The report covers its traffic history and IoCs.