San Diego (USA) Personal habits of customers, such as common times for banking transactions and even the way they hold their cell phones, have become the main tool for banks to prevent fraud in the face of the advancement of generative artificial intelligence, which can be used to defraud biometric systems.
This is because the technology that allows the creation of images, texts, and audios can replicate faces and voices, which would facilitate unauthorized access to bank accounts.
“These technologies can be manipulated, and we need to understand that,” says Scott Zoldi, director of Fico, which provides software to major banks in Brazil, during an event in the USA about innovations in banking technology.Money transfer on WhatsApp; through the analysis of customer behavior, banks seek new tools to combat fraud
– Gabriel Cabral/FolhapressWith each transaction or purchase on a credit card, various data are collected by the financial institution, in addition to the traditional values and recipients of the operation. Banks also collect information such as the time, location, and even the way the cell phone is held and the security steps are taken.”
[Bank security] will always be a layered defense. So, even though AI can steal my voice and my face, it is much more difficult to simulate me, and that’s where transaction data is really important,” says Zoldi, who holds more than 130 patents in data analysis and AI mathematical models.
“Any transaction the thief makes is not the one I would have made,” he adds.The more data banks have, the greater the chance they have of detecting fraud accurately, which can prevent bureaucracy and facilitate consumer usability without compromising security.
This process in banks also includes open banking, the sharing of data from the same client between institutions with prior authorization, and hyper-personalization, which is the ability to design individual banking products according to the needs and finances of each individual.”
These models predict whether a person can buy a specific item, if they have the money for it, if they would buy it, or if it deviates from their habits, and even if they could be being coerced. If the action deviates slightly from the pattern, it can already raise suspicion,” says Zoldi.Despite the constant advancements in banking technology,
analytical models and internal bank policies still present a number of so-called false positives, that is, the erroneous identification of fraud.According to Zoldi, correcting this type of flaw is a matter of sophistication and technological improvement.
“Over time, they will become increasingly better,” he says.New credit scoreAnother use for personal data is in the more assertive granting and collection of credit, according to experts in the sector.
With hyper-personalization and analysis of all possible information about each customer, banks are able to automate the design of personalized products according to the needs, payment capacity, and life stage of each customer.
This can be useful, for example, in offering real estate financing when the customer starts looking for property values, and in collection, with the most appropriate channel (WhatsApp, email, call, etc.) and time to do so.
With so much data on salary, expenses, and consumption habits, especially with open finance, financial institutions are also incorporating a new type of credit score, independent of those generated by traditional bureaus, with its methodology.
“Banks and financial institutions are not using the information in the way they should,” says Albert Morales, director of Belvo, the company that, in partnership with FICO, is behind the proposal for a new bank score, which is still in the testing phase.According to Morales, the new model will have immediate updates, unlike SPC and Serasa scores, for example.
“The normal score is slower. During the pandemic, it did not work because it takes time to be updated.
Also, checking your score lowers points in this traditional score,” he says.Increased credit card limitOne of the new possibilities expected to debut this year in the Brazilian market is the punctual granting of credit above the monthly card limit.
Imagine a situation in which an account holder is close to their limit at the end of the month but makes a purchase at the supermarket they usually frequent, at a routine value.
Instead of blocking this routine transaction, the bank allows the monthly bill to exceed the limit slightly, but without changing it for the following months.
Abnormal purchases would continue to be blocked.”Today, around 30% of disapproved credit transactions are for lack of a limit, but in all portfolios [of banks], there is a limit being granted that is not being used,” says Antonio Soares, president of Dock, the company that developed this new technology in partnership with FICO.
“This tool helps the bank distribute credit risk better,” he says.According to the executive, within six months, Brazilian institutions should start offering this novelty to customers.”
