Island Finds 186 Fake AI Domains Used in Typosquatting Attacks
Attackers mimicked 27 AI brands and used fake verification pages to push signed software and malicious Chrome extensions.
Security firm Island said it found 186 look-alike domains impersonating 27 AI brands, used in typosquatting attacks that rely on users mistyping the addresses of AI services, according to TechRadar.
Island said the fake sites profiled visitors' location, operating system and browser to serve different attack pages. One attack mimicked Cloudflare's verification screen and told users to copy and run a PowerShell command.
Another fake security-check page pushed a program called Net Guard, which carried a valid Microsoft Store signature but could change browser settings, install extensions and run JavaScript. Island also found malicious extensions that redirected search queries to third-party search services for ad revenue or data collection.
Most impersonated brands
ChatGPT was the most impersonated brand with 25 domains, followed by Claude with 18, DeepSeek with 16 and Runway with 12. Perplexity, Lovable and Character.AI each had 11 impersonating domains.
Of 30 malicious extensions Island found, 13 had been removed from the Chrome Web Store, while 17 were still listed as of the 3rd of this month.
Island advised users to check official addresses and bookmark the AI services they use frequently.
Quick answers
How many fake domains did Island find?
Island reported finding 186 look-alike domains impersonating 27 AI brands.
Which AI brand was impersonated the most?
ChatGPT was the most impersonated brand with 25 domains, followed by Claude with 18 and DeepSeek with 16.
What should users do to avoid these fake sites?
Island advised checking official addresses and bookmarking frequently used AI services.