Twitter says it removes over 1 million fake accounts a day

Twitter said it removes more than 1 million accounts considered spam every day from its platform, in a video conference with the press held this Thursday (7).

On the occasion, the company explained its policies to combat platform manipulation and the calculation of so-called monetizable users on a daily basis, those who are active on the platform.

Such data is important both from the point of view of combating disinformation and for those who advertise on the platform and seek real people as an audience. Fake user accounts on the social network were also targeted by Elon Musk. The billionaire questions Twitter’s figures that point out that spam accounts would represent less than 5% of the user base.

Last month, in a letter to Twitter, the Tesla owner said he could back out of the $44 billion deal to acquire the platform if the company did not provide requested data on spam and fake accounts. Before that, Musk had already made threats about the subject in publications on the social network itself.

As the company reported this Thursday, the platform’s anti-spam policies are focused on behavior, which ranges from more obvious spam content, such as those received by email, use of fake accounts and even sophisticated platform manipulation practices.

In the latter case, for example, a single real person may be operating a series of fake accounts in order to manipulate a particular debate.

The so-called bots, which consist of accounts that use automation to post tweets, are initially allowed by Twitter. The simple fact of using automation does not make an account considered spam, but the behavior of an eventual bot.

According to Twitter, despite the anti-manipulation policies employed by the platform, the company says it knows that it is impossible to remove all inauthentic accounts.

Over time, since the data was measured, the company’s quarterly calculations have maintained the estimate of less than 5% of spam accounts on the platform.

According to Twitter, this calculation is done through human reviewers who analyze random samples of users to verify whether or not they are spam.

It is this estimate that is the subject of questioning by Musk and has been used as an argument for the executive to eventually withdraw from the purchase.

​The company points out that the analysis cannot be replicated externally as it involves not only public information, but also personal data that only the company has access to.

In addition to the name and photo of users and frequency of tweets, for example, other data is used, such as IP address, phone number and other information about account activity on the platform.

In addition to removing spam accounts, the company said it also blocks millions of suspicious accounts weekly if they don’t pass checks by the platform to prove they are operated by people.

In 2018, the Washington Post reported that the company would be removing more than a million fake accounts a day. The platform, like other social networks, had been under pressure to act since Russian-operated disinformation campaigns emerged in the 2016 US elections.