Field report SE-76 / R5 · sheet /boards

Boards & GovernanceFiled from the territory

RBI Orders Board-Level API Security Overhaul at Banks, NBFCs

The RBI's 16 November 2026 advisory demands board-approved API policies, central inventories and AI-model safeguards after Rs 300 crore in losses from credential-theft attacks on lenders.

Grid
SE-76 / R5
Surveyed
November 20, 2026
Reading
3 min read
Correspondent
Amara Osei
API Security: RBI Sets Good Practices for Banks and NBFCs
Fig. SE-76 / R5API Security: RBI Sets Good Practices for Banks and NBFCs — blickpixel / pixabay

Field notes

  1. RBI advisory dated 16 November 2026 covers API governance at banks, NBFCs and urban co-operative banks.
  2. Estimated Rs 300 crore lost by smaller NBFCs in two months to API credential theft.
  3. Two Mumbai NBFCs lost about USD 2 million through more than 700 unauthorised transfers.
  4. CERT-In named the threat actor 'Crafty Falcon' on 4 November and warned the campaign was escalating.
  5. APIs connecting lenders to AI and machine learning models are covered for the first time.

India's central bank has made API security a board-level obligation for the country's lenders. The Reserve Bank of India (RBI) issued an advisory dated 16 November 2026, titled "Good Practices for API Governance and Security", directing banks, small finance banks, payments banks, NBFCs (non-banking financial companies) and urban co-operative banks to tighten how they build, expose, monitor and retire the application programming interfaces connecting them to fintechs, payment firms and AI services.

CNBC-TV18 and BW Businessworld reported the guidance on 17 November. It follows an RBI thematic review of API practices across regulated entities and goes wider than the regulator's earlier 3 November advisory on bank-NBFC fund-transfer APIs, covering every API across its full lifecycle and, for the first time, interfaces used to reach AI and machine learning models.

What does the advisory require?

The RBI stated the practices are illustrative and do not override existing legal or supervisory rules, but they set a clear benchmark for what supervisors will look for.

  • A board-approved API policy: a documented security framework covering internal APIs and those exposed to customers and partners.
  • Three lines of defence (a risk-governance model of operational control, oversight and independent audit): business and technology teams run controls, information security and risk oversee them, internal audit assures them.
  • Named owners for critical APIs: a business owner accountable for need and risk, a technology owner for build and operations.
  • A central API inventory covering partner-built, externally exposed and retired APIs, each classified by criticality and risk.
  • Least-privilege partner access: formal security reviews of third-party integrations, with access recertified periodically.

Which technical controls are specified?

The advisory calls for multi-factor authentication where appropriate and approved API gateways for critical and externally exposed interfaces. It also demands rate limits, quotas and throttling to curb misuse and denial-of-service attacks.

Requests and responses should be validated, while credentials, keys and certificates must be stored securely, rotated and monitored. Lenders must keep centralised API security logs with real-time monitoring for unusual access, data extraction and credential theft.

Security testing should go beyond software flaws, the RBI said, to check whether business logic — such as transaction approvals — can be abused through a legitimate API call.

Why are AI model APIs now in scope?

A notable addition covers APIs connecting lenders to AI and machine learning models. The RBI wants access controls matched to each model's sensitivity, safeguards against prompt- or input-based misuse, and audit trails of model versions, administrative changes and relevant inputs and outputs.

Data sent to outside model providers should be governed through data residency, minimisation, anonymisation or tokenisation (replacing sensitive values with non-sensitive substitutes), and retention rules, so customer data is not exposed through third-party AI services.

Why is the RBI acting now?

The advisory follows a run of attacks on lending and payment links. On 9 October, CERT-In (India's computer emergency response team) issued a critical alert on campaigns that stole API credentials of payment aggregators, gateways and partner banks, then called disbursement APIs from whitelisted systems to bypass KYC, sanction and limit checks. On 4 November it named the threat actor "Crafty Falcon" and warned the campaign was escalating.

Smaller NBFCs have been hit hardest, with an estimated Rs 300 crore lost in two months. Two Mumbai-based NBFCs reportedly lost about USD 2 million through more than 700 unauthorised transfers after the API link used for loan-disbursement instructions was compromised.

The regulator has also stepped up cyber checks on banks and has said technology resilience matters as much as credit risk.

What does it mean for lenders?

For CIOs and CISOs, the immediate task is visibility: a complete, risk-ranked list of every API and every partner connected to it. Institutions that rely heavily on fintech, co-lending and embedded finance partnerships will face the most work. The AI provisions mean GenAI pilots now require the same governance as payment APIs.

via bfsi.eletsonline.com (Source)

Share this article:

More from Amara Osei

Amara Osei

Show full bio

Market editor covering people, launches and day-to-day practice in corporate strategy and chief-executive leadership at CEO Frontier.

483 articles

Filed under

Adjacent sheets

« Previous article

End of sheetSE-76 / R5 · November 20, 2026