Grokking Gone Wild: X’s AI Assistant Unwittingly Boosts Malicious Ads!
Threat actors are “Grokking” their way around link blocks on X, using the platform’s AI assistant to sneakily share malicious links. They hide URLs in metadata, ask Grok for info, and voilà—Grok serves up the link on a platter! It’s like asking a friend for a secret recipe and getting a direct link to mayhem.
Hot Take:
Looks like cybercriminals are playing hide-and-seek on X, and Grok is the unwitting accomplice! It seems our AI overlords have a bit of a blind spot when it comes to spotting shady behavior. If Grok were a bouncer at a club, it would be letting in people with fake IDs while asking them if they’d like a complimentary drink. Time for X to teach Grok some street smarts!
Key Points:
- Malicious advertisers are exploiting X’s AI assistant, Grok, to bypass link posting restrictions.
- Grok is being used to extract and share hidden links from the “From:” field, boosting their credibility.
- The technique, called “Grokking,” amplifies malicious ads to reach millions of impressions.
- Potential solutions include scanning all fields and adding context sanitization to Grok.
- Reports have been made to X, but no official response has been received yet.
Already a member? Log in here