Hypertwish lets cons spear more phish

A security researcher has released an automated tool designed to launch sophisticated and targeted phishing attacks over Twitter.

The Hypertwish tool -- a play on the underlying mathematics in the app and so-called twishing attacks -- went far beyond the common automated phishing attacks where spammers blast out piles of malicious shortened URLs to bait victims before the offending account is banned.

Rather, it compiled and issued tweets based on trust and intelligent randomisation, issued shortened URLs designed to track victims, and exploited relationships between followers to build legitimacy.

"We need a better way of making fake profiles, automating tweets and exploiting trust," said Hypertwish creator and senior penetration tester at Booz Allen Hamilton, Sean Palka.

"You don't want to rely on mass fake URLs ... We only want to create accounts as needed. Some people create thousands of accounts and fall into the trap of [being] predictable".

The tool automated Twitter accounts with the command line client Twidge and crafted phishing tweets by swiping legitimate tweets and hashtags and using context-free grammar.

The latter syntax rules facilitated polynomial growth, which could make a small tweet highly randomised with the application of only a few rule sets.

All this was hidden from the user who saw an interface which simplified account following, tweet stealing, and the generation of coherent random tweets for baiting targets. The engine generated unique tweets when the user refreshed the page.

To make the con more believable, Palka built a graph by tweaking hyperbolic trees within the open source JavaScript InfoVis Toolkit.

This granted insight into follower relationships, allowing attackers to mention followers in tweets to increase the chance that targets would click malicious links.

Attacks were best placed where triangles were created in the tool's visual grid of relationships between Twitter accounts. Those indicated likely trusted relationships, which were useful to social engineers. 

"I can start sending communications to accounts referencing [their followers] and it will look like I know something about them. Or I can include all of them and it will look like I'm forwarding information that's relevant to them," Palka said at the Hack3rcon event this month.

The popular Maltego toolkit could be used to determine these relationships, but Palka said it was tough work to handle the necessary complex filters.

In a simple demonstration of the tool at the recent Defcon security conference, Palka created a fake account, which copied all tweets sent out by event staffer Ryan Clarke.

It was enough to get his fake account ranked higher than Clarke's, and even retweeted by Defcon organiser Jeff Moss, which scored him followers.

Hypertwish also kept an accurate list of victims who clicked malicious URLs via HTTP logs which parsed bots.

Palka found Twitter bots automatically requested URLs but did not dive deeper into the links to examine content.

Hypertwish took advantage of this by redirecting victims via iFrames to target payload sites, a move that went unnoticed by the bots.

The tool can be downloaded free for Linux.

Copyright © SC Magazine, Australia