Ask most Nepali bank customers whether their bank "uses AI," and many would say no. But open a banking app from Global IME Bank, NIC Asia, or Nabil Bank, and there's a good chance you've already chatted with an AI assistant, had a transaction silently screened by a fraud model, or had a digital loan application processed faster because a risk-scoring algorithm did the initial review. AI hasn't arrived in Nepali banking with fanfare — it's been rolled out quietly, feature by feature, inside apps people already use every day.
This piece walks through exactly where AI is being used in Nepal's banking and finance sector today, why Nepal Rastra Bank is now stepping in with formal guidelines, and what it could mean for the people who currently do the jobs AI is starting to help with.
How Nepali Banks Are Quietly Adopting AI
Nepal's banking sector has spent the last decade building up its digital infrastructure — mobile banking apps, QR payments, connectIPS, and digital wallets like eSewa and Khalti becoming part of daily life. AI adoption is following the same pattern: rather than one dramatic rollout, banks are layering machine learning quietly into systems customers already use. Global IME Bank's VIVA AI Assistant is one of the more visible examples, handling routine queries like KYC updates, password resets, and loan information. NIC Asia and Nabil Bank have been reported to be investing in similar AI-driven customer service and fraud-monitoring capabilities as part of their digital banking strategy.
Much of this adoption is happening behind the scenes in fraud monitoring, risk scoring, and compliance systems — areas where customers rarely see the technology directly but benefit from faster, safer transactions. Nepal's central bank has also nudged this shift along: in March 2025, Nepal Rastra Bank amended its online payment system directives, specifically encouraging payment companies to integrate AI and machine learning into their risk management systems for fraud detection.
Chatbots and Customer Service Automation
Chatbots are the most visible face of AI in Nepali banking today. Instead of waiting in a branch queue or holding on a call, customers can now ask an in-app assistant about their account balance, recent transactions, KYC document requirements, credit card details, or loan eligibility — often getting an instant, reasonably accurate answer. Global IME Bank's VIVA AI Assistant is a well-known example, built specifically to handle these routine, high-volume queries so human staff can focus on more complex requests.
For banks, this reduces the load on call centers and branch staff during peak hours. For customers, it means faster answers to simple questions, at any hour, without needing to visit a branch. The trade-off, as with chatbots everywhere, is that more complex or emotionally sensitive issues — a disputed transaction, fraud complaint, or loan rejection — still need a human agent who can exercise judgment the chatbot isn't designed for.
Fraud Detection and Cybersecurity Use Cases
Fraud detection is arguably where AI is delivering the most tangible value in Nepal's financial sector right now. Traditional rule-based fraud systems flag transactions based on fixed thresholds — for instance, any withdrawal over a certain amount. AI-based fraud detection instead learns typical behavior patterns for an account and flags transactions that deviate from that pattern in real time, catching more sophisticated fraud that static rules would miss, such as rapid small test transactions used to check if an account is restricted before a larger theft attempt.
This matters more than ever in Nepal, where digital payment fraud has grown alongside digital adoption. Scams involving fake investment or trading apps — including a widely reported case where a fraudulent "AI trading" app reportedly defrauded victims of hundreds of millions of rupees — highlight exactly why real, bank-grade AI fraud detection on the defensive side is becoming essential, not optional. Nepal Rastra Bank's own Financial Intelligence Unit continues to rely on suspicious transaction reporting from banks, and AI-assisted monitoring is increasingly part of how banks generate those reports faster and more accurately.
AI in Loan Approval and Credit Scoring
Nepal Rastra Bank's Digital Lending Guideline, first issued in 2022, already permits licensed banks and financial institutions to dispense loans entirely through digital platforms or licensed payment service providers acting as agents — removing the requirement of physical presence for many loan types. This digital-first lending approach is exactly where AI-based credit scoring adds value: instead of relying purely on manual paperwork review, banks can use automated models to assess a borrower's repayment ability and cross-check existing debt exposure with the Credit Information Bureau more quickly.
This can meaningfully speed up loan decisions and extend credit access to customers who might not have an extensive paper trail with a single bank. At the same time, it raises a legitimate concern: an opaque scoring model that quietly disadvantages certain borrowers is much harder to challenge than a human loan officer's decision. This is precisely the kind of risk Nepal Rastra Bank's newer AI-specific guidance is now trying to address directly.
Regulatory Challenges: Rastra Bank and Data Privacy
In December 2025, Nepal Rastra Bank released a draft of dedicated AI Guidelines for public comment — a significant step beyond general digital lending or payment rules, specifically targeting how licensed institutions should govern their AI systems. The draft applies broadly: commercial banks (Class A), development banks (Class B), finance companies (Class C), microfinance institutions (Class D), Nepal Infrastructure Bank, and payment system operators and providers like digital wallets are all in scope.
The guidelines cover AI use in credit scoring, fraud detection, customer service, risk management, and compliance monitoring, and place clear accountability on each institution's board of directors and senior management for the outcomes their AI systems produce. Boards are expected to define the institution's AI risk tolerance, set strategic direction for AI adoption, and build governance structures with real oversight — not just delegate AI decisions to the IT department. The draft also pushes for explainable AI: customers should be able to understand, in plain terms, why an AI system influenced a decision affecting them, with proper audit trails maintained behind the scenes. Data handling under these AI systems is expected to align with Nepal's existing Individual Privacy Act 2075, though independent reviewers have noted the guidelines are still fairly high-level and don't yet specify exact technical thresholds for bias testing or model validation — leaving room for uneven implementation across institutions until more detailed rules follow.
What This Means for Banking Jobs
AI adoption in Nepali banking is unlikely to eliminate large numbers of jobs overnight, but it is shifting what those jobs look like. Routine, repetitive tasks — answering basic customer queries, flagging obviously suspicious transactions, initial document verification for KYC — are increasingly handled or pre-processed by AI systems. This reduces demand for purely transactional customer service roles over time, but it's simultaneously creating new demand for people who can build, monitor, and govern these systems: fraud analysts who understand both banking risk and how AI models behave, compliance officers familiar with explainable AI requirements, and IT security staff focused specifically on AI system integrity.
For someone currently working in Nepali banking, the more strategic response isn't to fear automation outright, but to move toward the judgment-heavy, relationship-driven, and oversight-focused parts of the job that AI genuinely struggles to replace — handling disputed cases, complex lending decisions, and the human escalation path that NRB's own draft guidelines explicitly require banks to keep in place.
Frequently Asked Questions
Which Nepali banks are currently using AI?
Global IME Bank, NIC Asia, and Nabil Bank have been publicly reported to be using AI-driven chatbots, customer service tools, and fraud-monitoring systems as part of their digital banking strategy, with Global IME Bank's VIVA AI Assistant being one of the most visible public-facing examples.
Is there a specific law regulating AI use in Nepali banks?
Nepal Rastra Bank released draft AI Guidelines for public comment in December 2025, specifically targeting licensed banks, microfinance institutions, and payment service providers. As of early 2026, these remain in draft form rather than a finalized binding directive, but they signal the direction of upcoming regulation.
Can AI reject my loan application in Nepal?
AI-based credit scoring can influence a loan decision by assessing repayment ability and risk, but under Nepal Rastra Bank's draft AI guidelines, banks are expected to maintain human oversight and explainability for such decisions, meaning a fully automated, unexplained rejection would not align with the proposed regulatory expectations.
How does AI help with fraud detection in Nepali banking?
AI fraud detection systems learn typical transaction patterns for an account and flag unusual activity in real time, catching sophisticated fraud patterns — like small test transactions before a larger theft — that fixed, rule-based systems often miss.
Will AI replace bank jobs in Nepal?
AI is more likely to change banking roles than eliminate them entirely in the near term, automating routine customer service and initial fraud or document screening, while increasing demand for staff who can manage, audit, and make judgment calls around AI-driven systems.
Discussion