AI writing phishing texts and selling of the blocked accounts on Instagram

Check the latest news from the world of cybersecurity to be aware of the newest dangers of the cyberworld. This time the focus is on the scammers selling access to blocked Instagram accounts as well as the neural networks which have learnt to write very realistic phishing messages.

The access to a blocked account on Instagram can be purchased

One of the newest forms of fraud in social networks is related to Instagram. Scammers have created an innovative way of making the users of this popular service to pay.

How does it work?

The fraud scheme is extremely simple and yet, very efficient. First, scammers choose an Instagram account. Then, they create a clone of this account which looks quite realistic. Of course, they are using various personal information about this person, such as real photos, name and bio. The next step is simply complaining to the administration of Instagram claiming that the real person is actually a fake and this person should be blocked for stealing someone’s identity. Unfortunately, the system of identification on Instagram is not perfect and, the service really blocks real account leaving the fake ones behind.

How do the scammers earn on the scheme?

In fact, scammers get money in two ways. The first one is helping other scammers to ban real people. This function is not expensive and they earn some 60 USD for one account. Still, their major source of income is helping banned people to get unblocked. This assistance can cost even up to 5  000 USD.

Sometimes scammers refer to their victims right after the latter get their accounts clocked. Of course, they are offering their help in unblocking accounts.

Another way of getting your Instagram account blocked

Of course, a fake account is not the only account which is blocked by the administration of Instagram. There are many other ways in which scammers can make an account blocked and one of the most popular among them is reporting a person sending illegal material through Instagram. The administration of the service will block such accounts if there is a sufficient number of complaints. In order to make it happen, scammers are using a large amount of other accounts.

What is next?

The problem was reported by MotherBoard. Instagram has already reacted to this news encouraging users who come across such a problem to report it to the technical support of the service.

AI is more successful in composing phishing messages than real people

Undeniably, you are already aware of absolutely unbelievable things which can be achieved with the help of artificial intelligence. Neural networks have proved to be successful in solving various tasks. This time the researchers of the Government Technology Agency of Singapore checked the ability of neural networks to create phishing messages. The algorithm used for this purpose was not just successful. It turned out to be more effective in addressing this problem than real people.

How was the research carried out?

The researchers from Singapore first created a hundred of phishing messages on their own. Then, they used the algorithm for natural speech analysis and other machine learning algorithms for processing their texts. After training the algorithms, they used them for composing similar content. Both the messages created by the researchers and the messages written by artificial intelligence were sent to 200 employees of the institution. This was done in order to assess which of the messages look more realistic.

Instead of links to fraudulent websites, the messages were just equipped in trackers the aim if which was checking to which messages the employees would react.

The results of the research

In fact, the team of the researchers did not assume the possibility of people finding the messages composed by machine learning algorithms more realistic. Yet, the results of the research showed exactly this trend. People tend to click on the links located in the messages composed by AI more frequently than on the links in the messages created by real people.

One of the most amazing things the AI system did while creating its messages was adding extra elements in a quite creative way. For instance, knowing the recipients of the messages would be from Singapore, it successfully added citations from the laws of this country for increasing the realism of the messages.

Will real scammers use this technology?

This question does not have one answer. On the one hand, as the research has shown, neural networks are extremely successful in creating phishing messages. On the other hand, this technology is still rather expensive for the commercial usage while massive spamming with phishing mail is highly cost-effective without the help of AI.