How artificial intelligence and other emerging technologies can prevent bank fraud
State-run Punjab National Bank (PNB) has been brought to its knees by a few of its own bankers. Between the various acts of omission and commission, it appears that a small island hosting renegade bankers can go undetected even when surrounded by an ocean of honest bankers. One is reminded of a famous battlefield saying: “We have met the enemy and he is us.”
Bill Gates, the billionaire co-founder of tech giant Microsoft, once said that in the future we will need banking, but not necessarily banks. JPMorgan Chase & Co. chief executive Jamie Dimon went on record to say that a bank is but a technology company in disguise. Perhaps the time is ripe to ask: Even as we need more banking, do we need more bankers?
Can modern technology run banks with more efficiency and less fraud than humans? A modern aircraft such as the Dreamliner flies 98% on autopilot and 2% on the manual mode. Statistics suggest that commercial air travel today is safe, reliable, and largely pro-customer. Why can’t banking be the same?
Wherever and whenever financial transactions involve valuable assets, the possibility of fraud increases exponentially. High levels of risk arise when there are too many people involved, or there is a lack of transparency in processes and visibility to data, or an awful lot of manual effort is required to execute transactions. On such fertile ground, frauds take root -- in the form of collusion between internal and external stakeholders.
The PNB fraud is a classic example where a small number of employees manually executed lots of illegitimate transactions to benefit a specific customer.
Today, we are at a defining moment where emerging technologies can provide a new bedrock for trustful computing. Trustful computing can significantly reduce financial frauds, human irregularities, and gross misadventures in financial deals and transactions. Though fraud may never entirely vanish, it can be made highly improbable.
There are three technological building blocks that financial institutions need to urgently embrace: Complete digital integration, anomaly detection driven by artificial intelligence, and blockchain. Let us examine each of them individually.
Complete digital integration involves conversion of all artefacts into digital form, and joining up islands of automation in a seamless way. All paper-based transactions must be eliminated. For example, letters of understanding, letters of credit, or collateral such as securities and real estate must be captured as digital assets accessible in a secure way.
Commercial transactions involve a consortium of producers, traders, distributors, bankers, insurance providers, lawyers, agencies, and more. Since every transaction is only as strong as its weakest link, the consortium itself needs to be on a federated, interconnected platform, or it is a time bomb waiting to go off.
The absence of integration at PNB between the core banking system and Swift (Society for Worldwide Interbank Financial Telecommunications) is a case in point. The need of the hour is extreme, real-time digital integration, and free information flow across the consortium based on explicit rules.
Anomaly and fraud detection using artificial intelligence and machine learning is just as critical. Historically, financial institutions have treated fraud as rare unfortunate accidents to be dealt with after the event. This approach is now thoroughly outdated.
Anomalies must be treated as ongoing incidents to be actively managed by harnessing the power of automation. It is now established that insiders, no matter how few, play a role in most financial frauds. Therefore, the old saying, “trust, but verify”, comes to mind. While institutions should continue to place trust in their people and invest in them, they must also allow machines to verify that the institutional trust is not being abused by anyone.
Automation and artificial intelligence can be used to actively monitor high-frequency transactions. These emerging technologies can help detect anomalies in transactions, user interactions, account management and change in asset status using fraud-detection techniques, sentiment analysis, irregularity checks and spotting of outliers.
Finally, blockchain is an idea whose time has come. It guarantees incorruptible and immutable transactions without human intervention or oversight. In the world of blockchain, transactions are stored in online ledgers that allow recording of new transactions but do not permit anyone to change or delete old transactions.
Moreover, copies of the ledger are distributed to and actively maintained by every stakeholder within the consortium. It is the consortium that can collectively decide and verify the authenticity of each transaction automatically. There is no human intervention in this entire process of transaction verification. If any stakeholder in the consortium rejects the transaction, it is not written to the ledger at all.
In short, blockchain enables the automation and assurance of trust between stakeholders with diverse interests, which is so important for banking today.
In terms of fraud, PNB may not be an isolated case, and we must proceed with the assumption that if one bank is vulnerable, so are many others.
Banking is ultimately a business of trust. We believe the time is ripe for financial institutions and regulators to subscribe to emerging technologies that can reduce fraud, strengthen governance, and enhance customer trust.
Just as necessity is the mother of invention, let this current crisis be the mother of change. As British wartime prime minister Winston Churchill said: “We must never let a good crisis go to waste.”
Santanu Paul is the managing director and chief executive and Ritesh Modi is lead faculty, blockchain, at TalentSprint Pvt. Ltd, a Hyderabad-based skill development and training company.