
Imagine a vault with a high-tech lock but a spare key hidden under the doormat. That’s essentially the situation facing many of India’s smaller Non-Banking Financial Companies (NBFCs). While they rushed to digitize lending—approving loans in minutes and disbursing cash instantly—their cybersecurity budgets and governance didn’t keep pace. Most of the lenders hit in the recent wave of attacks have annual turnovers of around ₹1,000 crore or less, operate with small security teams, and rely on long audit cycles. This makes them "softer targets" compared to large banks that have dedicated security operations centers and deeper pockets.
The core issue isn’t sophisticated malware in the traditional sense. It’s basic hygiene gaps. Officials point out that some small NBFCs don’t even know what infrastructure they should have or when audits should happen. They often rely on third-party fintech firms to host their websites and lending apps on cloud platforms like Amazon Web Services (AWS). This creates a critical blind spot.
Here’s the loophole that’s draining bank accounts. When an NBFC hosts its systems on the cloud, its firewall typically "whitelists" the IP addresses of that cloud provider’s data centers. The logic is simple: only allow payout requests from these trusted addresses. But here’s the catch—cloud services are publicly available. Anyone can open an account on AWS and get an IP address from the same pool.
Fraudsters exploit this by registering their own cloud accounts, writing code that mimics the NBFC’s disbursement instructions, and sending payout requests from these whitelisted IPs. To the NBFC’s bank, the request looks legitimate because it comes from a trusted address. The system that was supposed to pay borrowers ends up paying hackers instead. In the case of Hero FinCorp, attackers used at least 165 different IP addresses to execute 317 transactions, siphoning off about ₹12.84 crore. This technique bypasses the firewall entirely, allowing the money to flow out instantly.
The attack chain usually starts with compromising public-facing systems. CERT-In advisories describe attackers exploiting exposed user acceptance testing (UAT) servers, file-upload flaws, or deprecated APIs to gain initial access. Once inside, they hunt for the crown jewels: payment gateway and partner bank API credentials. These are often found hard-coded in source code or stored in plain text configuration files—a cardinal sin in security.
Armed with these valid credentials, attackers don’t need to hack the core banking system. They simply call the disbursal API directly, instructing the bank to transfer funds. Crucially, these calls skip the NBFC’s own application-level checks like loan creation, KYC verification, and beneficiary validation. The money lands in "mule accounts"—bank accounts held by money mules—and is quickly moved further through the network or converted to cryptocurrency like USDT before vanishing overseas.
What’s new is the speed and scale. Attackers are reportedly using AI-driven offensive tools like CyberStrikeAI, ARTEX, and PentAGI to automate the entire process. These tools can autonomously execute attack chains, generate thousands of tailored scripts, and adapt their tactics based on the target’s responses—all at "machine speed".
In one October case, police found that a Chinese AI platform named CyberStrikeAI was used to send command scripts to an NBFC that lost ₹3 crore. This automation lowers the barrier to entry. As one official put it, "One just needs to put the toolkit in the AI model and give a prompt, and the agent will do the hacking for you". The result is a compressed timeline between finding a vulnerability and cashing out, often shorter than the victim’s remediation cycle.
The human cost of these technical failures is stark. Hero FinCorp discovered the fraud when HDFC Bank sent system alerts about unauthorized debits. By then, ₹12.84 crore had left through 317 transactions routed to 34 different bank accounts. Police arrested five individuals, including a bank relationship manager, and managed to place a lien on about ₹1.7 crore. The investigation revealed links to Chinese Telegram groups and the conversion of funds into USDT.
Unifinz Capital India provides another grim example. The company reported that on October 20—a Dussehra holiday—₹7.69 crore was transferred through 346 transactions in just over two hours. The payments bypassed normal authorization procedures and were only detected the next morning during daily reconciliation. A few days later, the company reported a second incident involving ₹7.27 crore across 309 transactions. While police figures cite a loss of ₹12.49 crore via 550 transactions for Unifinz, the company’s disclosures suggest the total impact could be higher, highlighting the confusion that often accompanies these breaches.
A critical vulnerability is operational, not technical. Many smaller NBFCs reconcile their loan accounts only weekly or fortnightly. They only realize a customer never received the loan days or weeks after the money has been stolen. By then, the trail is cold. The "golden hour"—the crucial window immediately after a fraud when funds can be frozen in mule accounts—is long gone.
This delay significantly impacts recovery rates. In the Delhi-NCR region alone, 12 NBFCs lost over ₹65 crore in three months. Police managed to detect and seize about ₹18 crore before it left the banking system, but the rest is likely gone. Nationally, reports estimate losses of around ₹300 crore across roughly 150 complaints, though this figure remains unconfirmed by regulators. The relationship between IP usage and loss size isn’t linear—Hero FinCorp’s 165 IPs facilitated a ₹12.84 crore heist, but the damage is driven more by the duration of the attack and the lack of real-time blocks than the number of addresses used.
The Reserve Bank of India (RBI) has clearly had enough. Following the attacks, the regulator intensified scrutiny, asking banks to prove they are implementing existing cybersecurity rules and to audit their API links with NBFCs and fintechs. A confidential advisory issued on November 3, 2026, mandates that banks require NBFC clients to undergo security audits by CERT-In-empaneled auditors.
New requirements are strict. Banks must enforce beneficiary whitelisting or "maker-checker" approvals for payouts by December 3, 2026, and complete audits of existing NBFC partners by February 3, 2027. They are also required to monitor API transactions for anomalies like volume spikes, odd-hour activity, and bursts of new beneficiaries. The message is clear: outsourcing technology does not outsource accountability. Banks are responsible for every rupee that moves through their rails, even if the instruction comes from a partner’s system.
For NBFCs, the path forward requires significant investment. Global benchmarks suggest cybersecurity should consume 12-13% of IT budgets, with financial services ideally allocating 0.8-1.0% of revenue. Smaller NBFCs, which have historically under-invested, now face a double whammy: the direct cost of implementing controls like secrets managers, multi-factor authentication, and 24/7 monitoring, plus the potential cost of lost business if partner banks deem them too risky to onboard.
Strategic investments must focus on closing the specific gaps exposed by these attacks. This includes retiring hard-coded API tokens, isolating development and production environments, and implementing real-time anomaly detection that can spot a ₹1 test transfer followed by a massive payout. The use of AI by attackers means static defenses are no longer enough; NBFCs need behavioral baselines for every API client to detect deviations instantly.
The long-term impact on the NBFC sector will be profound. Cybersecurity is transitioning from a technical cost center to a board-level operational risk. We can expect a period of consolidation where smaller, weaker players either invest heavily to survive or get acquired. Profitability may take a hit as compliance costs rise, but the alternative—reputational damage and massive fraud losses—is far worse.
The ₹300 crore wake-up call has arrived. For NBFCs, the choice is stark: upgrade the digital locks and remove the spare keys, or risk leaving the vault door open for the next AI-powered heist.