As mobile financial fraud becomes increasingly sophisticated, new intelligent systems that track how users type and swipe provide a powerful shield for loyal customers.
Imagine a college student in Dhaka preparing to pay his final semester tuition fees or a small business owner calculating his monthly income in a busy market. Suddenly, you receive an urgent call from someone claiming to be a customer support executive from a trusted mobile banking provider. The person on the phone seems very professional and warns me that unless I verify my details immediately, my account will be permanently blocked due to a sudden system upgrade. Panicked and impatient, they follow instructions without thinking. Within minutes of a seemingly innocuous interaction, a student’s entire tuition fee and a business owner’s hard-earned profits can be completely wiped out. These scenarios are no longer isolated nightmares, but reality for many across the country. Fraudsters have evolved far beyond simple trickery, deploying sophisticated social engineering tactics to bypass standard security measures and financially harm the public before they even realize what went wrong.
The mobile banking landscape in Bangladesh has experienced tremendous growth over the past decade. Platforms like bKash, Nagad and Rocket have fundamentally changed the concept of financial inclusion, with more than 144 million registered users entering the formal economy as of January 2026, of which 570,000 are relatively vulnerable youth accounts, according to Bangladesh Bank data.
Increased internet penetration and widespread use of smartphones have shifted everyday transactions to digital screens. However, this large-scale change has also attracted highly organized fraud syndicates. As reported by the Daily Star in May 2024, a total of 48,586 personal mobile financial services accounts were suspended by the Bangladesh Financial Intelligence Unit (BFIU) for suspected involvement in online gambling, gambling and hundi.
Scammers are siphoning millions of Tk from unsuspecting users through fake investment schemes, cloned emergency numbers and highly coordinated social engineering tactics. As transaction volumes soar to hundreds of billions daily, the financial and mental burden on everyday users is increasing. This growing epidemic directly threatens the core trust necessary for a thriving digital economy and poses a critical national issue that requires immediate intervention.
The main vulnerability that enables these crimes lies in the way current security systems operate. Traditional bank defenses rely heavily on strict rules-based methods that simply monitor for obvious red flags, such as multiple incorrect PIN entries or unusually large and uncharacteristic transfers. Unfortunately, today’s sophisticated fraudsters rarely use brute force to hack systems.
Instead, they manipulate victims into willingly sharing their one-time passwords or logging into applications in completely standard ways using stolen credentials. These criminals so carefully mimic legitimate login procedures that traditional security rules cannot distinguish between real account holders and remote thieves. If the PIN matches and the one-time password is correct, the system blindly assumes the transaction is secure and processes the theft.
To combat this rapidly evolving threat, our research introduces a smarter and more adaptive framework. We focused on a new concept: behavioral biometrics. The concept works on the simple but powerful principle that the way you type, swipe, and scroll on your phone’s screen is as unique to you as your physical fingerprint. When you combine this continuous behavioral data with transaction patterns such as where you are, when you typically send money, and how much you typically trade, a very comprehensive behavioral profile emerges.
The development of this solution involved intentional advancements in machine learning models. We initially utilized autoencoders to closely profile normal user behavior. They then moved to advanced networks that can capture time-based sequences, applied gradient boosting techniques, and finally combined these elements to build a robust ensemble system that can learn from vast amounts of data.
The performance results of this hybrid research work are very promising for the future of mobile security. Our system achieved an impressive 97 percent fraud detection rate and 95 percent accuracy. To put this improvement in perspective, running only the initial baseline model would have missed 67 percent of fraud. This huge jump in accuracy means that the system not only catches more criminals, it also catches them with incredible accuracy. A high accuracy rate has an important practical advantage of reducing false alarms. This ensures that loyal customers no longer face frustrating delays or unexpected account blocks while attempting to make legitimate payments. While analyzing the framework, we found that the most important metric for discovering anomalies was a direct combination of the user’s geographic location and the user’s specific scrolling and typing speed.
For Bangladesh, adopting this kind of intelligent framework could be completely transformative. Regulators like Bangladesh Bank, along with leading mobile financial service providers, have an immediate opportunity to integrate these predictive models directly into their existing digital infrastructure. The framework is designed to be highly adaptable and locally relevant, providing real-time, deployable defenses against certain social engineering tactics currently prevalent in the ecosystem. Such proactive security mechanisms are absolutely essential to securing the next stage of the nation’s journey towards a truly cashless society.
Securing the digital economy requires a rapid shift from reactive troubleshooting to proactive artificial intelligence-driven defense. It is essential that regulators, traditional banks, and fintech companies work together to invest in advanced behavioral protections. By adopting these intelligent systems today, the financial sector can finally outpace fraudsters and ensure that digital financial services remain a safe and empowering tool for all citizens.
Shuvashish Roy is a Senior Research Fellow in the Research & Innovation Division of Prime Bank PLC and Md Tuhin Rana is a student in the Department of Statistics, Dhaka University.
