Cyber Risks · The Hacker News

Cybercriminals Exploit X’s Grok AI to Bypass Ad Protections and Spread Malware to Millions

Summary: Researchers at Guardio Labs documented a technique they call "Grokking," in which malvertisers run promoted video posts on X with adult content as bait and hide a link in the unscanned "From:" metadata field below the video player. The scammers then tag Grok from throwaway accounts asking where the video came from, prompting the AI assistant to display the link publicly under a promoted post that drew hundreds of thousands of impressions, directing users to fake CAPTCHA scams and information-stealing malware.

Bottom line: Guardio found hundreds of accounts each posting hundreds or thousands of such posts over a few days, staying active until suspended, and the link's appearance in a Grok reply also boosted its SEO and domain reputation - meaning the platform's own trusted AI account became the distribution channel your users were taught to trust.

Read the original at The Hacker News ↗

Topics
AI Security
Related