Toss Bank's AI system spots fake IDs, prevents fraud

Toss Bank logo / Newsis

By Anna J. Park

Toss Bank announced that its real-time artificial intelligence (AI) system has contributed to preventing financial fraud. Since launching earlier this year, the digital bank said its AI-based detection system has spotted over 600 fake IDs during the past six months, undermining attempts to open up fraudulent accounts.

The bank explained that its ID detection system can determine in real time whether an ID photo submitted by a new customer opening an account is an authentic image or not.

If the system detects that the information in a submitted ID photo does not match the customer's information or if it is a re-photographed image from a smartphone screen or a printed ID, all financial transactions, including the opening of the account, are restricted until re-verification is completed. The system's accuracy rate is reported to exceed 94 percent, the internet-only bank said.

Toss Bank began developing the AI detection system in May last year and underwent field tests in October the same year. The system was officially launched in February this year. Over the past six months, the real-time detection system has restricted financial transactions in 600 fake ID cases, and is estimated to have prevented some 12 billion won ($8.9 million) worth of potential fraud.

“As an internet-only bank, it is paramount to develop our own technology and capabilities that guarantee safe services, such as online face identity verification procedures,” a Toss Bank official said. “The bank will continue to make significant investments in developing AI and data-related technologies to enhance its ability to provide customers with secure financial services.”