
23 Oct 2026
An Indian lender's guide to compliant EMI reminders, recovery-calling restrictions, borrower protection and AI vendor accountability.
AI loan collection calls India must follow the RBI requirements applicable to the lender, loan type and purpose of contact. For overdue-loan recovery covered by RBI's August 2022 circular, calls are prohibited before 8 a.m. and after 7 p.m. A separate recovery window applies to microfinance loans.
Calling within permitted hours is not enough. Lenders must also address harassment, borrower confidentiality, outsourcing, recording, telecom rules and human escalation.
Before deploying automated collection workflows, banks and NBFCs need to establish what the AI may say, when it may contact borrowers, and who remains accountable when something goes wrong.
RBI does not impose a blanket prohibition on automated loan-reminder technology in the regulatory materials examined. However, AI loan collection calls India must be assessed against the rules governing the underlying activity.
Three interactions require distinction:
Interaction | Purpose | Regulatory consideration |
| EMI reminder | Notify borrowers before an instalment becomes due | Servicing, data and telecom requirements |
| Overdue follow-up | Contact borrowers after a missed payment | Applicable recovery-conduct restrictions |
| Debt recovery | Seek repayment of outstanding dues | Recovery-agent rules, borrower protection and lender oversight |
An overdue recovery call does not become ordinary servicing simply because the AI describes it as a reminder.
The RBI circular dated 12 August 2022, RBI/2022-23/108, addresses the responsibilities of regulated entities engaging recovery agents.
It covers specified commercial banks, co-operative banks, NBFCs, asset reconstruction companies and All India Financial Institutions. Microfinance loans covered by the separate RBI microfinance framework are excluded from this circular.
The circular prohibits threatening or anonymous calls, persistent harassment, privacy intrusion, inappropriate messages, intimidation and misleading statements.
For AI loan collection calls India, lenders must ensure these restrictions are reflected in approved scripts, scheduling and escalation processes.
Recovery activity | Applicable restriction | Source |
| Overdue recovery covered by the August 2022 circular | No calls before 8 a.m. or after 7 p.m. | RBI/2022-23/108 |
| Overdue microfinance recovery | Calling before 9 a.m. or after 6 p.m. is treated as a harsh practice | Microfinance Directions, 2022 |
| Pre-due-date servicing | No universal recovery-specific window can be inferred | Assess relevant servicing and telecom requirements |
The RBI microfinance FAQ clarifies the special recovery restriction.
These are distinct regulatory rules, not interchangeable calling windows. A lender must also consider frequency, contact preferences and circumstances, even when a call occurs during permitted hours.
RBI recovery agent calling rules apply to the conduct of the underlying collection activity, whether communications are handled by people or technology.
Appropriate behaviour | Prohibited or unacceptable behaviour |
| Identify the lender and purpose | Anonymous or threatening recovery calls |
| Communicate verified amounts | False balances or misleading consequences |
| Respect contact restrictions | Persistent harassment and prohibited out-of-hours calls |
| Verify identity before discussing dues | Disclosing debt to unauthorised third parties |
| Provide assistance channels | Humiliation, intimidation or abuse |
| Escalate disputed payments | Pressuring borrowers using unverified demands |
For AI loan collection calls India, scripts should use lender-approved information instead of generating unsupervised payment threats, legal warnings or settlement commitments.
A polite synthetic voice does not make a coercive collection practice acceptable.
The regulated lender retains responsibility for outsourced activities under applicable RBI requirements.
The RBI Digital Lending Directions, 2025 define a Lending Service Provider (LSP) by reference to its activities, potentially including digital-loan servicing, monitoring and recovery.
A software supplier is not automatically an LSP or recovery agency. Classification depends on the functions performed and contractual arrangement.
Area | Lender | AI service provider |
| Compliance | Determine applicable rules | Disclose services performed |
| Scripts | Approve communication | Implement approved conversations |
| Calling hours | Set required restrictions | Demonstrate scheduling controls |
| Loan information | Maintain accurate records | Process authorised information |
| Complaints | Own investigation and resolution | Route and document exceptions |
| Audits | Establish oversight requirements | Provide agreed records and reports |
| Security | Approve data-handling controls | Follow contractual safeguards |
This is an illustrative operating model, not a universal regulatory allocation.
Before approving AI loan collection calls India, lenders should determine whether the vendor merely provides software or performs outsourced servicing or recovery.
AI EMI reminder calls should clearly identify the lender, protect account information and offer assistance without unnecessary pressure.
Illustrative script
AI assistant: "Hello, may I speak with [borrower's name]?"
Borrower: "Speaking."
AI assistant: "I'm an automated assistant calling on behalf of [lender name] regarding your loan account. May I complete a brief identity check before discussing any details?"
After successful approved verification:
"Our records indicate an EMI of ₹[amount] due on [date]. Would you like repayment information or assistance from our support team?"
This is not an RBI-prescribed script. The lender must approve authentication, recording notices and wording.
For AI loan collection calls India, automated assistants should not request passwords, PINs or OTPs through unsafe authentication methods, invent penalties or promise unauthorised repayment concessions.
An effective collections workflow must recognise when automation should stop.
Borrower response | Recommended action |
| "I've already paid." | Refer for payment reconciliation |
| "The amount is incorrect." | Register and escalate the dispute |
| "I've lost my job." | Offer authorised hardship assistance |
| "This isn't my loan." | Stop account disclosure and investigate |
| "Stop calling me." | Record the request and review permissible future contact |
| "I want a person." | Transfer or provide the approved support channel |
| Borrower reports serious distress | Stop collection pressure and initiate sensitive-case escalation |
For AI loan collection calls India, escalation should be based on the borrower's response, not simply whether a payment commitment has been captured.
AI should not independently approve loan restructuring, discretionary settlements or legal actions without appropriate authorisation.
Following February 2026 draft proposals, RBI issued final recovery-conduct amendments in August 2026 for multiple regulated-entity categories.
The commercial-bank amendment and NBFC amendment specify 1 January 2027 commencement.
Their recovery provisions address governance, recovery-agency oversight, borrower communications, grievances and documentation.
Covered institutions must assess requirements relating to recovery-call time and frequency records, recording of specified call content or text, and preservation for six months, subject to applicable exceptions for court proceedings.
For AI loan collection calls India, preparation should distinguish existing obligations from provisions that commence in 2027. Institutions must consult their own category-specific directions rather than applying commercial-bank provisions universally.
AI debt collection compliance includes borrower-data handling, recordkeeping and protection against unauthorised disclosure.
A lender should evaluate whether its workflow can document:
A text transcript should not automatically be assumed to replace an audio recording where the applicable direction requires particular call records.
For AI loan collection calls India, privacy assessment must consider the Digital Lending Directions and India's phased data-protection regime.
Read Sicada's DPDP Act and AI voice agent compliance guide for related background. Its legal statements should be read against the phased commencement of the DPDP Act and Rules, including major substantive provisions scheduled for May 2027.
Yes. RBI recovery rules and telecom communication regulations must be assessed separately.
TRAI's Telecom Commercial Communications Customer Preference regulations and amendments address commercial communications, including automated calling requirements.
TRAI also issued a 19 November 2025 direction on mandatory 1600-series adoption for service and transactional calls by specified regulated financial institutions.
Institution category | Adoption deadline |
| Commercial banks | 1 January 2026 |
| Large NBFCs above ₹5,000 crore in assets, Payments Banks and Small Finance Banks | 1 February 2026 |
| Remaining NBFCs, co-operative banks, RRBs and other specified entities | 1 March 2026 |
For AI loan collection calls India, lenders must verify call classification, originating-number arrangements and applicable autodialler or robocall requirements.
Adding loan cross-selling to an EMI reminder may introduce promotional-communication obligations. Calling regulations should therefore be reviewed for the actual message, not merely the campaign's name.
Before launching AI loan collection calls India, lenders should complete this implementation review.
Requirement | Pre-launch question |
| Regulatory scope | Which RBI directions cover the lender and loan? |
| Call purpose | Is the interaction servicing or recovery? |
| Calling restrictions | Are applicable hours and contact controls enforced? |
| Script approval | Has compliance reviewed all conversation branches? |
| Borrower privacy | Are sensitive details disclosed only after appropriate verification? |
| Vendor role | Has the LSP/recovery-agency classification been assessed? |
| Human assistance | Can disputes and hardship cases reach authorised staff? |
| Telecom compliance | Are sender, number-series and automated-calling obligations addressed? |
| Records | Can required information be retrieved and preserved? |
| Oversight | Who investigates failures and approves remediation? |
The checklist supports implementation planning. It does not certify compliance.
Sicada.ai provides AI-powered voice communication technology. Its published service terms describe voice workflows and related capabilities, while explaining the customer's responsibilities concerning telephony providers and applicable regulations.
For AI loan collection calls India, lenders can evaluate an approved workflow involving borrower contact, verified repayment information, response capture and human escalation.
Before deployment, required features such as lender-system integrations, call-hour enforcement, recording retention and audit exports must be confirmed and tested for the actual implementation.
RBI does not impose a general ban on AI-assisted calling in the materials discussed. Legality depends on the lender, loan, calling purpose and applicable regulatory obligations.
The August 2022 RBI circular restricts covered overdue-loan recovery calls after 7 p.m. Microfinance recovery has a separate restriction after 6 p.m. Later applicable directions must also be considered.
Not necessarily. Pre-due-date servicing and recovery of overdue loans are different activities, although other telecom, privacy and servicing obligations may apply.
Only within authorised lender procedures. Discretionary settlements, restructuring and sensitive disputes require appropriate oversight.
Not automatically. The lender must evaluate the precise recording, documentation and retention requirements applicable to its institution.
AI loan collection calls India should be deployed as controlled borrower-contact workflows, not unrestricted automated recovery campaigns.
Lenders must identify the applicable RBI rules, enforce recovery-calling restrictions, protect borrower information and maintain oversight of outsourced service providers. TRAI requirements and upcoming RBI amendments also require attention.
A credible deployment should answer four questions: What can the AI say? When can it call? When must it escalate? And what evidence can the lender produce if a conversation is challenged?
Discuss AI-Assisted EMI Reminder Workflows
Explore lender-approved reminder scripts, borrower-support pathways and human-escalation options with Sicada.ai.
Products
Resources
Others
All rights reserved. Powered by Edysor