Vulnerabilities / Threats

10/23/2018
04:38 PM
Connect Directly
Twitter
LinkedIn
Google+
RSS
E-Mail
50%
50%

Battling Bots: How to Find Fake Twitter Followers

Duo researchers explain the approach they used to detect automated Twitter profiles and uncover a botnet.

The discovery of a massive botnet can start with finding a few fake Twitter followers, report Duo researchers investigating the process of identifying and analyzing automated accounts.

Earlier this summer, Duo's Olabode Anise, data scientist, and Jordan Wright, principal research and development engineer, wrapped up a project investigating how they could detect Twitter bot characteristics. The goal was to create a means of differentiating automated and legitimate accounts, and they built a classifier tool to distinguish bots based on a pre-defined set of traits.

Their research dug into one of the largest random datasets of public Twitter accounts to date, they report. As part of their project, Wright and Anise identified three specific types of bots serving different purposes. "Content-generating" bots actively create new content (spam, malicious link), "amplification" bots like and retweet content to boost a tweet's popularity, and "fake followers" are a type of amplification bot intended to inflate users' popularity.

The two today published a new report digging into the latter. Their analysis covers how fake followers operate, how they discovered an initial list of fake followers, and how they leveraged that list to unearth a botnet made up of at least 12,000 Twitter accounts.

"We understand the fake followers are just as important to the social ecosystem," says Anise. "They artificially inflate the ratio of followers to followees."

On one hand, fake followers can be used to harass, or compromise the credibility of, legitimate accounts. On the other, they can boost the popularity of fake accounts, making them appear more credible than they are. The researchers gave a talk at Black Hat USA discussing how a botnet spoofed legitimate accounts to evade detection and spread a cryptocurrency scam.

Spot the Bots

It's tough to tell when a follower is fake, and researchers explain more information is better when analyzing accounts. In general, fake followers are hard to detect on an individual level because they don't show much activity – aside from, of course, following other accounts.

But a lack of activity doesn't mean an account is malicious. Some people create Twitter accounts simply to follow other users and stay current on the news, Anise explains.

So instead of hunting fake followers on an individual basis, the researchers decided to consider their full social networks. Fake followers are typically purchased and used as groups; as a result, they tend to share characteristics because they are developed by the same operator.

But which traits set fake followers apart from real ones? After Anise and Wright wrapped their initial pool of research, the botnet they were watching began to use fake followers to trick people into thinking spoofed Twitter accounts were real. They took a closer look at the followers of a fake Elon Musk Twitter profile and explored their similarities.

"One thing that we saw was, [they were] pretty easy to identify," says Anise of the fake followers. "These bots weren't really trying to hide … if you notice patterns or have a similar account, you can use it to pivot and find other bot accounts."

Timing is one key factor. If a large group of accounts suddenly follows the same profile, for example, there's a higher likelihood they're fraudulent.

Accounts following the fake Elon Musk profile had a proverb or fortune in their profile description – a quick and easy means of bypassing spam detection. Profile completion is an easy way to determine the quality of a bot; if an account has a profile, it appears to be real. However, creating unique profiles is harder than generating random usernames.

With this in mind, the researchers could separate these bots from legitimate followers. Because they were studying the fake accounts as a group, they could observe whether similar accounts had similar behaviors. Once they found a small group of fake followers, they could branch outside that network and look for other fake accounts with similar traits.

How to Crawl for Followers

Anise and Wright uncovered the botnet with a "one-degree crawl" of a single fake follower. They found a fake account and looked at their social network, as well as the social network for each account the fake follower was following. They then applied a script to search the social network for a specific account. The result is a web of fake and legitimate profiles, connecting which fake accounts are following legitimate ones.

While it's a good start, not every bot in a botnet will follow the same people, meaning the researchers may not have caught entire groups of fake followers. To find new ones, Anise explains, they can use a bot found during their initial crawl and search its network for new fake followers. After doing this, they uncovered an additional 1,200 bots.

The two point out that large groups of fake followers have patterns that are easier to recognize; smaller groups, in contrast, may be more subtle. To find smaller groups of fake accounts, the researchers first determined when multiple accounts were created on the same day and consecutively followed a target account.

Related Content:

 

Black Hat Europe returns to London Dec 3-6 2018  with hands-on technical Trainings, cutting-edge Briefings, Arsenal open-source tool demonstrations, top-tier security solutions and service providers in the Business Hall. Click for information on the conference and to register.

Kelly Sheridan is the Staff Editor at Dark Reading, where she focuses on cybersecurity news and analysis. She is a business technology journalist who previously reported for InformationWeek, where she covered Microsoft, and Insurance & Technology, where she covered financial ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
neiljakson105
0%
100%
neiljakson105,
User Rank: Guru
11/7/2018 | 11:06:24 AM
nice post
Great article. I totlly liked it. Your entire website is great. I am a finance brokers. We got folks awesome mortgages and other commercial finance. buy dissertation online
muneebkhatri
100%
0%
muneebkhatri,
User Rank: Apprentice
10/28/2018 | 5:47:50 AM
dissertation writing service
This was among the best posts and episode from your team it let me learn many new things.  
Joe Stanganelli
50%
50%
Joe Stanganelli,
User Rank: Ninja
10/26/2018 | 10:33:21 PM
Fake unique profiles
> "However, creating unique profiles is harder than generating random usernames."

Sure, but maybe not that much harder.

Just throw a bunch of self-designations and cliches into a program, and have it randomly spit out a few.

Example: "Marketing enthusiast, dad of 3, professional cat herder, Giants fan. Tweets are my own."
SchemaCzar
0%
100%
SchemaCzar,
User Rank: Strategist
10/24/2018 | 10:28:22 AM
I thought they just declare every conservative account is fake
ooops, I guess that will get me booted...
aronsalsa
50%
50%
aronsalsa,
User Rank: Apprentice
10/23/2018 | 5:19:48 PM
Fake twitter sccount
We all know that how social media is impacting everyone nowadays and twitter is one of the biggest platforms to make any statement to the world. There are many fake twitter accounts also and that has to be recognized and the given details here really making a point of out of that. 

Dell error code 2000-0142 is really supporting that.
Veterans Find New Roles in Enterprise Cybersecurity
Kelly Sheridan, Staff Editor, Dark Reading,  11/12/2018
Understanding Evil Twin AP Attacks and How to Prevent Them
Ryan Orsi, Director of Product Management for Wi-Fi at WatchGuard Technologies,  11/14/2018
7 Free (or Cheap) Ways to Increase Your Cybersecurity Knowledge
Curtis Franklin Jr., Senior Editor at Dark Reading,  11/15/2018
Register for Dark Reading Newsletters
White Papers
Video
Cartoon
Current Issue
Flash Poll
Online Malware and Threats: A Profile of Today's Security Posture
Online Malware and Threats: A Profile of Today's Security Posture
This report offers insight on how security professionals plan to invest in cybersecurity, and how they are prioritizing their resources. Find out what your peers have planned today!
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2018-19355
PUBLISHED: 2018-11-19
modules/orderfiles/ajax/upload.php in the Customer Files Upload addon 2018-08-01 for PrestaShop (1.5 through 1.7) allows remote attackers to execute arbitrary code by uploading a php file via modules/orderfiles/upload.php with auptype equal to product (for upload destinations under modules/productfi...
CVE-2008-7320
PUBLISHED: 2018-11-18
** DISPUTED ** GNOME Seahorse through 3.30 allows physically proximate attackers to read plaintext passwords by using the quickAllow dialog at an unattended workstation, if the keyring is unlocked. NOTE: this is disputed by a software maintainer because the behavior represents a design decision.
CVE-2018-19358
PUBLISHED: 2018-11-18
GNOME Keyring through 3.28.2 allows local users to retrieve login credentials via a Secret Service API call and the D-Bus interface if the keyring is unlocked, a similar issue to CVE-2008-7320. One perspective is that this occurs because available D-Bus protection mechanisms (involving the busconfig...
CVE-2018-19351
PUBLISHED: 2018-11-18
Jupyter Notebook before 5.7.1 allows XSS via an untrusted notebook because nbconvert responses are considered to have the same origin as the notebook server. In other words, nbconvert endpoints can execute JavaScript with access to the server API. In notebook/nbconvert/handlers.py, NbconvertFileHand...
CVE-2018-19352
PUBLISHED: 2018-11-18
Jupyter Notebook before 5.7.2 allows XSS via a crafted directory name because notebook/static/tree/js/notebooklist.js handles certain URLs unsafely.