Sberbank has developed a service called Aura, which recognizes fraudsters during a telephone conversation and immediately warns users about it, the bank’s press service reported.
The technology was created on the basis of artificial intelligence: a language neural network model of deep learning investigated several thousand conversations of real attackers. In the framework of the pilot project, the system showed an efficiency of 80%. Moreover, in less than 1% of cases, the model was mistaken.
At the moment the service is available to Sbermobile subscribers. You can connect it yourself. If the artificial intelligence considers that the client is talking to an attacker, it will give an audio message: “Attention! This call is probably fraudulent. ” At the same time, “Aura” learns on the basis of new material in real time.
Dmitry Teplitsky, CEO of ABK, which participated in the development of the technology, noted that in the first half of 2021, more than 70% of calls from unknown numbers came to customers from fraudsters. According to him, every month with the help of social engineering they steal from the citizens from 3.5 billion to 5 billion rubles. At the same time, criminals are constantly inventing new schemes.
The technical director of Sbermobile, Evgeny Sitnikov, said that so far the service has been launched in about 12 regions of Russia, but in the future it will expand its geography, TASS reported. The system will operate in pilot mode until the end of the year. There are currently no plans to extend the technology to other operators.
In early September, Stanislav Kuznetsov, deputy chairman of the board of Sberbank, said that the bank had collected a “library of voices” from telephone scammers, with the help of which it monitors calls to customers and blocks the withdrawal of funds from accounts. Prior to this, the credit organization collected a database of fraudulent numbers. As of the end of April, it had over 600,000 contacts.
.