• Jaime González Gasque

How do banks and fintech companies use voice recognition to prevent fraud?

Updated: Jun 23, 2021


  • Voice biometrics makes it so easy for financial institutions to verify the identity of their customers, while improving the user experience.

  • A system that includes voice recognition also allows the creation of voiceprint records, to identify possible scammers.

  • Financial education and user awareness about the security provided by these tools is key for the sector.


Where there is money, there are also people willing to get hold of it through deception. Bank fraud is historically a game of cat and mouse, a new layer of security is followed by a new way of circumventing it; and as new forms of attack are identified, technologies are developed or actions are implemented to prevent, stop and / or track them. Fintech companies have also joined this race, which have the perfect model to face the digital trend of consumers but face the great challenge of the entire financial industry: offering the client the security, certainty and confidence that their resources are safe , while providing the convenience of technology to run their operations and working to make financial services easier, more accessible, faster, and without much paperwork. Data security and protection are not trivial topics; In 2020 alone, the National Commission for the Protection and Defense of Users of Financial Services (Condusef) received 5 million 866,000 complaints for cyber fraud, an increase of 2.6 times in 5 years. The same organism indicates that 70% of the frauds to users of financial institutions were made by cybernetic means; this has almost doubled in the last 4 years.


What are the new ways to verify your financial identity?


Over the past 3 decades, numerous ways to confirm a user's identity have emerged, coexisting and even combining in today's ecosystem. The first and oldest is to ask the user for something that (theoretically) only they know, for example a password or PIN; through something they own, such as tokens, which generate a unique key, special cards or their own smartphone; And more recently, taking advantage of advances in cybersecurity tools and advances in artificial intelligence (AI), it can be done through something that makes them unique: fingerprints, facial recognition or voice. Digital channels allow more agile and faster inquiries and transactions. In such an environment, to prevent and repel fraud, the ideal is a system that uses a combination of different tools; in a short time biometric mechanisms will be the norm and we will even see the combination of several of them. The advantage that we see in voice biometrics is that, from a simple sample —which can be taken in a branch, through a call or an application—, a software is in charge of performing the authentication and current technology allows a double layer of security. First the identification of the voice under unique patterns analyzed by artificial intelligence, which validates hundreds of variables that make our voice unique and unrepeatable, even among members of the same family or people who have similar voices and, second, machine learning, applied by example in random questions, to avoid the use of a recording or a person with a similar voice, but without information that the user has. This enables a user-friendly and quick-to-implement security tool for fintech and banking. But the use is not limited to authenticating, the use of virtual voice assistants allows you to make inquiries, request operations and even make purchases. In addition, there are additional layers of security, through machine learning, these systems can receive extra help, by detecting and alerting inconsistent behaviors according to a profile created from previously collected information, in addition notifications and measures can also be automated, without requiring control calls, and allowing the user to verify their identity or authorize transactions via automated systems.


What are the advantages of speech recognition in the financial system?

Among the resources that have been implemented within the sector is machine learning-assisted speech recognition, which has some intrinsic advantages:

  1. It requires just a microphone, which is easier to find on a device (or even works with a traditional phone) than a fingerprint scanner or facial recognition. The experience improves significantly for the user since they do not have to remember any data, have a specific device or simply slow down the processes.

  2. Avoid using a password "12345" or simply leaving your password written down on a post-it, and thereby jettison any company security measure.

  3. The financial institution can establish multiple verification moments that when carried out in the background do not become a nuisance, it is more if the transaction is done with voice commands, even the financial company can use this interaction to do the validation.

  4. A system that includes voice recognition also allows the creation of voice fingerprint records, which makes it possible to identify fraudsters, who can be easily identified to block suspicious operations or activate the appropriate protocol.

  5. It seems trivial, but either as an add-on or as the primary authentication mechanism, it is not affected by clothing or accessory obstructions.

  6. How many times have you had to remove your mask because you cannot authenticate in your online banking app? Or in a cold winter city, have you had to remove your gloves in order to try to make a transaction and authenticate your fingerprint?

Mobile banking, as one of the digital financial services most used by users in Mexico, is an area where there are some of the great opportunities and challenges for AI-assisted cybersecurity. It is very interesting, for example, what Crediorbe de Colombia or Kubo Financiero de México are doing, which are implementing voice AI in their collection processes. Financial education and user awareness about the security provided by these tools, in addition to their operational benefits to streamline routine processes, such as balance inquiries, payment authorization and transactions, are an ongoing topic for the industry. In Mexico, it was recently made mandatory for mobile banking users to share their location with financial institutions to access their accounts, in order to prevent fraud. Although the measure has been controversial and not to everyone's liking, it is an indication that the sector is taking action on the matter and that there are still processes to improve, where they must allocate a good part of their investment if they really want to create locks strong to protect users. The same investment in cybersecurity represents a reduction in costs for the companies themselves. By sophisticated fraud prevention systems, processes are automated that reduce response time to users, eliminate cumbersome procedures and reduce the work required for human intervention. If all these filters can be harmoniously combined (voice recognition, finger registration, geolocation, etc.) for a seamless experience, we will be taking an important step to provide users not only with real security, but with the perception of it. they need to make a more mature and complex use of digital banking and the financial alternatives offered by fintech to manage their money, investments, credits and other financial products. Humberto Pertuz is an engineer by profession, entrepreneur and passionate about new voice technologies.


He is the CEO and founder of Vozy, a developer of voice solutions based on artificial intelligence for the Spanish-speaking market. The opinions published in this column are the responsibility of the author and do not represent any position on the part of Business Insider Mexico. Discover more stories at Business Insider México

10 views