Vulnerabilities in bank chatbots allow hackers to steal money

Awillix specialists discovered vulnerabilities in bank chatbots that could allow fraudsters to transfer money without the knowledge of customers. Positive Technologies confirmed the risks. The largest banks reported that they limit the functionality of chatbots in messengers. 

It should be noted that about 10% of Russian banks use chatbots: they can be used in messengers, mobile applications, social networks, on the website and in the contact center.

Alexander Gerasimov, Director of Information Security at Awillix, said that chatbots in messengers, which are used for individual account transactions, may be vulnerable to malicious attacks.

The company’s specialists checked the security of chatbots in two Russian credit organizations and found similar logical vulnerabilities. They allow obtaining the number and expiration date of cards, as well as finding out the account balance and cell phone number of the client.

“During the pentests, it was possible to log into the test client’s account and perform a money transfer operation,” Alexander Gerasimov said.

Maxim Kostikov, head of the banking systems security research group at Positive Technologies, confirmed that chatbots can be subject to various vulnerabilities, which depend on their functionality. For example, security problems can allow you to get customer data, get into their personal accounts in the chatbot, and find out the card balance.

According to him, the most popular scenarios of deception are changing the functionality of the chatbot to collect information about the person who uses it, sending malicious software on behalf of a credit institution, replacing the robot with a fraudster during communication, creating fake chatbots of banks. 

“If a person uses a bank chatbot, which is able to make money transfers in the messenger, two-factor authentication can be configured to log into the application to protect funds,” stressed Infosystems Jet expert, adding that there is also a danger in cases when an attacker gained direct access to the victim’s device physically or as a result of a malicious attack.