Cybersecurity researchers have discovered a sophisticated new technique being used by cybercriminals to bypass X’s malvertising protections and distribute malicious links with the help of the platform’s AI assistant, Grok.
The technique, codenamed “Grokking,” was detailed by Nati Tal, head of Guardio Labs, in a series of posts on X. It circumvents restrictions on Promoted Ads, which typically allow only text, images, or videos, by using a clever combination of metadata and AI amplification to reach massive audiences.
According to Guardio Labs, attackers run video card-promoted posts containing adult-themed bait content. Instead of embedding malicious links directly in the post, which would be blocked, they hide them in the “From:” metadata field under the video player — a field that X apparently does not scan for prohibited links.
Once the bait post gains traction, the attackers tag Grok in the replies with questions like “where is this video from?” This triggers the AI assistant to fetch and display the hidden link in its response, effectively legitimizing it.
“Adding to that, it is now amplified in SEO and domain reputation – after all, it was echoed by Grok on a post with millions of impressions,” Tal said.
“A malicious link that X explicitly prohibits in ads (and should have been blocked entirely!) suddenly appears in a post by the system-trusted Grok account, sitting under a viral promoted thread and spreading straight into millions of feeds and search results!”
Guardio Labs said the links lead to malicious ad networks that redirect users to fake CAPTCHA scams, information-stealing malware, and other dangerous content through smartlink monetization techniques.
The domains have been linked to a Traffic Distribution System (TDS), a tool frequently used by malicious ad tech actors to funnel unsuspecting users toward harmful or fraudulent websites.
Researchers have identified hundreds of accounts using this method, each posting thousands of times until they are suspended for policy violations. “They seem to be posting non-stop for several days until the account gets suspended for violating platform policies,” Guardio noted, calling the operation highly coordinated and large-scale.
This discovery raises serious concerns about the potential abuse of AI assistants like Grok for malicious campaigns, especially when combined with paid promotion tools.
