Hordes of Fake Users Scamming Social Media, eCommerce Sites

By Mike Elgan  |  Posted 2016-07-03 Print this article Print
Fake Followers

Even real accounts can be converted to fake accounts. The practice of hijacking a large number of legitimate accounts and turning them into fake accounts seems to be getting more common. Individuals or companies with large social followings are taken over and replaced by bots, usually bots that automatically post material objectionable to the targeted account.

For example, the Twitter accounts used by thousands of comedians, athletes, journalists and others this year have been taken over by spambots, blasting adult content to people who had followed each celebrity.

In the wake of the mass shooting in Orlando, Fla., which took place in a gay nightclub and for which the terrorist group ISIS claimed responsibility, hackers cracked the Twitter accounts of various ISIS sympathizers and auto-posted Gay pride tweets. Some of those ISIS accounts themselves were already fake accounts created to illegally support and recruit for ISIS, and became doubly fake after the hack.

As with all social networks, Facebook has been plagued by fake user accounts. Some are created to scam people and con them out of money. Other fake accounts are set up to create a cover of anonymity for the harassment and trolling of real users.

Facebook is alone in asserting a so-called "real names" policy, but has no way to enforce it. If your actual name sounds unusual—say, Rusty Pipes or Amanda Blackhorse—Facebook may demand to see proof. But if anyone creates a fake account using a more common-sounding name, Facebook won't question it. Facebook's "real names" policy is really a "normal names" policy.

Recently, Facebook started using an automated system to flag possible fake accounts, which are sent to a team of Facebook employees for review.

The Creation of Fake Followers Will Become an AI Arms Race

The fake-user problem is a large, damaging and growing one.

TeleSign research found that 82 percent of businesses struggle with fake users and that 43 percent deliberately allow them to exist in their "ecosystem" because taking steps to weed them out would annoy legitimate users during the registration process. The TeleSign report estimated that 10 percent of all user accounts are fake and that 21 percent of real users have been victimized in one way or another by these fake accounts.

The problem of fake user accounts has gotten so bad that some countries, including the UK, are now thinking about prosecuting perpetrators criminally. Those cases would be limited to the creation of fake accounts for the purposes of trolling and harassment.

Sadly, law enforcement will never be able to keep up with the stealthiness, speed and volume of fake-user account economy. We need better technology.

A Cornell computer science graduate student named Yixuan Li and a professor named John Hopcroft are working with Google to figure out how to spot "fake social engagement." They point out that it's cheap and easy to buy thousands of fake followers on Twitter, YouTube, Amazon and Facebook and then use those fake accounts to simulate user engagement. The purpose of all that engagement is to game social site algorithms to boost the rankings of a video or post.

The researchers' system is called "LEAS" (Local Expansion At Scale). The idea is to create something called an "engagement relationship graph" of accounts that are known spamming accounts, and watching for groups of accounts that work together in ways that are unlikely to happen among ordinary users.

They've so far focused on YouTube, but the method can be extended to massive data sets and other social networks.

LEAS is a good start, but won't be able to keep up with future trends in the faking of followers.

One of the most significant trends in the past two years is the availability of new artificial intelligence tools for developers. Big data AI is rapidly becoming a commoditized, ubiquitous service—you know, like electricity or Internet connectivity.

Fake user account spambots will become so good and programmable through AI that it will be impossible to tell which accounts are real and which are fake.

The coming wave of artificially intelligent bots will create fake accounts by the millions, which will turbo-charge harassment, astro-turfing, trolling, fraud and propaganda.

The bottom line is that if you think your social media friends and followers are a bunch of phonies, you may be exactly right.



Submit a Comment

Loading Comments...
Manage your Newsletters: Login   Register My Newsletters

Rocket Fuel