By : Mr. Neehar Pathare, MD, CEO & CIO, 63SATS Cybertech
Technology has made life more convenient, but it has also made it easier for fraudsters and cybercriminals to track and attack. Cybercriminals have become faster, forcing financial organisations to simultaneously strengthen their defence systems as well.
It’s no longer a concern of whether the next significant bank hack will take place but rather when it will occur and, more critically, whether advanced technology can instantaneously intervene to avert it. Conveniently, the rapid and transformative advances made using artificial intelligence and machine learning suggest that the answer is increasingly leaning towards a resounding yes.
Historically, cybersecurity has taken a defensive approach, an unending struggle of responding to events after they occur. Investigations and patch deployments only occur after a breach takes place. Such a stance is insufficient when facing foes who not only have greater resources but also use automation on their side. It is estimated that cybercrime will cost the world about $15 trillion by 2029, which surpasses even the GDP of many countries. Even banks and financial institutions face enormous losses, with cyber fraud claiming an estimated $42 billion a year, a 23% rise from 2020.
A New Frontier in Defence
Smart technology, especially advanced AI and ML, is evolving from mere tools into active combatants in this digital war. These aren’t your rudimentary rule-based systems that flag suspicious activity based on predefined parameters. Today’s smart tech, especially agentic AI, is autonomous. It perceives its environment, analyses massive streams of data, makes context-aware decisions, and acts—all while learning and refining its performance. Think of it not as a vigilant guard but as an ever-learning, always-on digital sentinel.
The effectiveness of AI in detecting fraud has moved beyond speculation. Recent studies have found that modern AI-driven fraud detection systems have a detection rate between 87% and 94%, simultaneously lowering false positives by 40% to 60% relative to older systems. As a result of fewer incorrectly flagged transactions, this change enhances customer experience and reduces bank costs. AI, for example, can monitor credit card use in real time and flag suspicious changes in spending habits, geo-location, or something as subtle as the device used for previous logins. It also detects unusual patterns, such as logins from different devices and mid-session credential check failures, which may indicate an account takeover.
Think about the enormous volume of financial transactions. Institutions handle over 1.7 trillion transactions each year. Trying to manually filter through this vast data for identifying fraudulent activities is like searching for a needle in a blindly navigated haystack. On the other hand, AI technology has the capability of processing and analysing thousands of payments per second and instantly evaluating each new transaction against prior records, user profiles, IP data, and geographic locations to flag red alerts.
The impact of AI is even greater. The combined AI techniques using both supervised and unsupervised learning are better than other methods at finding weaknesses in fraud detection systems and keep improving to deal with new threats. Their adaptability is especially important considering an ever-evolving landscape of potential cybercriminal threats. For example, they project cyber fraud losses in India to reach ₹1.2 lakh crore (about $14.5 billion) by 2025, which is nearly 0.7% of its GDP.
What really makes smart technology valuable is how proactive and adaptive it can be. When dealing with a zero-day exploit – a previously unknown vulnerability – traditional tools may struggle. However, an intelligent AI system that can spot unusual system behaviour or sudden API problems can isolate the affected systems, take temporary actions, predict possible ways the exploit could be used before the official diagnosis of the vulnerability, and neutralise threats before they escalate into full-blown incidents.
Challenges and Future
Achieving seamless AI integration, however, poses unique challenges. Data quality and availability remain paramount; AI models demand vast amounts of high-quality, labelled data; if historical data doesn’t adequately represent emerging fraud techniques, they can struggle to generalise.
The momentum cannot be overlooked despite the challenges standing in the way. The adoption of fraud-detecting AI technology has surpassed 50% among financial institutions by 2022. The financial sector is now leveraging AI to detect and prevent fraud in real-time, safeguarding customer interaction points, maintaining compliance with changing regulations, addressing evolving threats, and responding to emerging cybersecurity risks. All these enhancements are fundamental in building a more robust financial ecosystem.
In summary, advanced threats posed by bank hacks can be neutralised with the right technology—in this case, the rise of AI as an agentic autonomous system. Evolving from a passive to an active, adaptive defence posture enables instant detection and response to fraud, as well as breach prediction and risk mitigation with unparalleled precision and speed.
Protecting and cementing confidence in our digital economy involves more than raising security perimeters; cunning adversaries can be outsmarted and outmanoeuvred by deploying self-learning guardians that operate on a higher level of intelligence.