What AI governance obligations do the IRDAI Bima Sugam Regulations 2024 create for product ranking algorithms?

The IRDAI (Bima Sugam – Insurance Electronic Marketplace) Regulations, 2024 (F. No. IRDAI/Reg/5/199/2024, dated March 20, 2024) establish India's insurance digital public infrastructure marketplace. Regulation 10 mandates that the marketplace's algorithm for displaying and ranking insurance products must not create conflicts of interest — the ranking AI cannot systematically favor products from insurers with commercial arrangements with the marketplace operator over products that are genuinely better suited to the customer. Four specific AI governance obligations follow: (1) The product ranking algorithm must be documented with its ranking criteria — a black-box ranker that optimizes undisclosed commercial signals violates Regulation 10's fair conduct requirement; (2) Conflict-of-interest audit: the algorithm must be audited for features that correlate with insurer commercial relationships (advertising spend, commission arrangements, exclusive tie-ups) rather than product quality and customer suitability metrics; (3) Explainability: for any product recommendation surfaced to a customer, the marketplace must be able to explain why that product was ranked above alternatives — this is a customer-facing explainability obligation, not just an internal documentation requirement; (4) IRDAI examination readiness: the algorithm documentation must be available for IRDAI inspection under the Corporate Governance for Insurers Regulations 2024's board accountability framework.

Do AI chatbots deployed on Bima Sugam require an IRDAI insurance intermediary license?

Yes, if the AI chatbot solicits or procures insurance. The IRDAI (Insurance Intermediaries) Regulations classify entities that 'solicit or procure' insurance as intermediaries requiring IRDAI registration — as corporate agents, insurance brokers, or web aggregators depending on the nature of the activity. An AI chatbot deployed on Bima Sugam that guides a customer through product selection, answers questions about specific policies, recommends coverage amounts, or facilitates the purchase workflow is functionally soliciting or procuring insurance. The chatbot's operator is the licensed intermediary, and the chatbot's AI model outputs are attributable to the licensed entity. This creates three AI governance obligations for Bima Sugam chatbot deployments: (1) The chatbot's recommendation logic must satisfy the suitability assessment obligation applicable to the intermediary category — corporate agents are required to assess customer insurance needs; an AI chatbot that skips suitability assessment to shortcut to a product recommendation creates regulatory non-compliance for the licensed operator; (2) The chatbot's recommendation outputs must include the disclosures required by IRDAI Digital Distribution Regulations 2024 — AI involvement disclosure, basis of recommendation, key terms and exclusions, human advisor offer before purchase; (3) The chatbot must be able to document its recommendation basis for IRDAI examination — the AI's recommendation log is part of the licensed intermediary's records.

What AI governance obligations apply to premium pricing and underwriting AI on Bima Sugam?

Insurers deploying AI-driven premium pricing and underwriting models through Bima Sugam face obligations at two levels. First, IRDAI's product approval process: all insurance products sold on Bima Sugam must be IRDAI-approved. The approved product's premium rates are actuarially certified. An AI model that dynamically adjusts premiums at the individual customer level — using real-time behavioral data, telematics, or health data — may be deviating from the IRDAI-filed product rates in ways that require separate regulatory approval. Insurers must document whether their AI-generated individualized rates fall within the IRDAI-approved rate band or constitute a new product requiring separate filing. Second, the IRDAI Corporate Governance for Insurers Regulations 2024 require board-level accountability for product pricing decisions. An AI model making premium pricing decisions must have board-approved model governance, documented model inputs and outputs for actuarial certification, and audit trails that demonstrate the pricing outputs are consistent with IRDAI-filed rates. The actuary signing off on the product's pricing cannot substitute a generic actuarial certification for documentation of the specific AI model's pricing logic and validation results. The AI model and its outputs must be auditable at the individual policy level for IRDAI examination.

How does Bima Sugam's India Stack integration create a dual DPDP and IRDAI KYC compliance obligation for AI?

Bima Sugam is integrated with India Stack — Aadhaar for identity verification, DigiLocker for document sharing, and UPI for premium payment. This creates a dual-compliance layer for AI models processing customer data through Bima Sugam: (1) DPDP Act 2023 obligations: customer identity data fetched from Aadhaar and DigiLocker during insurance purchase on Bima Sugam is personal data (and potentially sensitive personal data — health data for health insurance, biometric data for Aadhaar verification). DPDP Act Section 6 requires purpose limitation — data fetched for insurance purchase cannot be used to train AI models for cross-selling, behavioral profiling, or any purpose beyond the stated insurance transaction; (2) IRDAI KYC Master Circular obligations: IRDAI's KYC requirements for insurance policyholders apply on Bima Sugam. AI models processing e-KYC data (V-CIP, OTP-based Aadhaar) must comply with IRDAI's KYC data quality and accuracy obligations. The intersection creates a gap that is unique to India's DPI-integrated insurance marketplace: an AI model that uses Aadhaar-fetched data for real-time underwriting risk scoring is simultaneously subject to DPDP Act purpose limitation (the data was consented for KYC verification, not risk scoring), IRDAI KYC Master Circular data governance, and IRDAI's product approval framework (if the risk scoring deviates from IRDAI-filed rates). No global insurance AI governance framework addresses this three-way intersection.

AI Governance · IRDAI · Bima Sugam · Insurance Distribution · India BFSI

IRDAI Bima Sugam Insurance Distribution AI Governance: Marketplace Algorithm Obligations, Chatbot Suitability, and AI Premium Pricing Under the 2024 Regulations

India's Bima Sugam — the world's first national insurance digital public infrastructure marketplace — launched under IRDAI Regulations 2024. AI recommendation algorithms, chatbot policy advisors, and AI-driven premium pricing on Bima Sugam face governance obligations distinct from the IRDAI WG-AI framework. Product ranking conflict-of-interest rules, chatbot intermediary licensing, actuarial certification of AI pricing outputs, India Stack dual DPDP+IRDAI KYC obligations. Zero global competitor coverage.

Zytra Research Team · October 2026 · 12 min read

📖 Part of the India BFSI AI Governance Complete Guide 2026 — RBI, SEBI, IRDAI, NPCI & DPDP in one place.

Bima Sugam, launched under the IRDAI (Bima Sugam – Insurance Electronic Marketplace) Regulations, 2024 (F. No. IRDAI/Reg/5/199/2024, dated March 20, 2024), is India's national insurance marketplace built on Digital Public Infrastructure principles — a unified digital platform through which all insurance products from all IRDAI-regulated insurers can be discovered, compared, purchased, and serviced.

Every IRDAI-regulated insurer must participate in Bima Sugam. Every insurance product sold in India can be sold through Bima Sugam. The marketplace integrates with Aadhaar for identity, DigiLocker for documents, and UPI for premium payment. At this scale, the AI models operating within Bima Sugam — ranking products, powering chatbot advisors, driving individualized premium pricing — are AI models operating at the center of India's insurance distribution infrastructure.

This post maps the AI governance obligations that the Bima Sugam Regulations and the supporting IRDAI framework create for each AI function in the marketplace. These are distinct from the IRDAI WG-AI framework (which governs insurer-level AI) — they apply specifically to AI on the marketplace platform and to AI distribution through it.

Product Ranking Algorithm: The Conflict-of-Interest Obligation

Bima Sugam's Regulation 10 establishes that the marketplace must operate in the interest of policyholders and must not create conflicts of interest in product presentation. The product ranking algorithm — the AI that determines which products appear first when a customer searches for term life cover, a health plan, or a motor policy — is the primary commercial lever of the marketplace.

An AI product ranker that systematically elevates products from insurers that pay higher fees to the marketplace operator, or that optimizes for click-through or purchase conversion rather than customer suitability, violates Regulation 10's fair conduct standard. The specific AI governance obligations:

Ranking criteria documentation: The algorithm's ranking criteria must be documented — what features drive a product's rank? Premium competitiveness, claim settlement ratio, product feature completeness, customer rating, or commercial factors? Documentation allows IRDAI to assess whether the ranking is consistent with Regulation 10.

Conflict-of-interest feature audit: The algorithm must be audited for features that correlate with insurer commercial relationships. A model that learned, through training on historical click data, to upweight products from insurers with higher advertising spends (because those products had more marketing-driven historical clicks) has embedded a commercial conflict even without an explicit commercial feature in the model design.

Explainability for customer-facing ranking: When a customer asks why a particular product was recommended over a cheaper alternative, the marketplace must be able to provide an explanation grounded in the customer's needs and the product's characteristics — not in ranking algorithm weights.

AI Chatbot Distribution: Licensing and Suitability Governance

AI chatbots on Bima Sugam that guide customers through product selection and facilitate purchase are, in regulatory terms, soliciting or procuring insurance. IRDAI's intermediary regulations classify this activity as requiring a license — as a corporate agent, insurance broker, or web aggregator. The chatbot operator is the licensed entity, and the chatbot's outputs are attributable to that licensee.

This creates a governance structure that maps AI outputs to regulatory obligations:

Suitability assessment AI: The chatbot's need-assessment conversation (questions about coverage needs, existing policies, family situation, budget) is the AI's suitability assessment. This assessment must be documented — a log of the chatbot's questions and the customer's responses, mapped to the products that were recommended, stored for the period required by the intermediary category's records retention obligation.

Mandatory disclosures in AI output: IRDAI Digital Distribution Regulations 2024 require specific disclosures before and during insurance purchase: AI involvement disclosure, basis of recommendation, key terms and exclusions, human advisor offer before policy purchase. The chatbot's output must include all these disclosures at the correct points in the customer journey. An AI model that has been trained to minimize friction in the purchase funnel by abbreviating or omitting disclosures is non-compliant with the Distribution Regulations regardless of purchase conversion performance.

IRDAI WG-AI adversarial testing: The IRDAI WG-AI cybersecurity guidelines specify adversarial testing for customer-facing AI in distribution channels — specifically prompt injection risk, where a malicious user could manipulate the chatbot into recommending unsuitable products, disclosing confidential insurer pricing information, or bypassing mandatory disclosures. Adversarial testing documentation must be part of the licensed operator's AI governance records.

AI Premium Pricing: The Actuarial Certification Gap

AI-driven individualized premium pricing creates a regulatory gap specific to India's product approval architecture. IRDAI approves insurance products with filed premium rate schedules. The approved rates represent the actuarially certified pricing for that product. An AI model that dynamically adjusts premiums at the individual customer level — using behavioral data, telematics, or health data — may be generating rates that deviate from the IRDAI-filed rate band.

If the AI pricing outputs are within the approved rate band (applying approved loading factors for individual risk characteristics that were disclosed in the product filing), they are compliant. If they generate rates outside the approved band based on AI-learned patterns not disclosed in the product filing, they constitute a deviation from IRDAI-approved pricing that may require a product amendment or separate filing.

The AI governance obligation: insurers must document whether their AI pricing model's output range falls within the IRDAI-approved rate band, and the actuary certifying the product must review the AI model's pricing logic — not only the aggregate pricing statistics. A generic actuarial certification that covers the product overall without reviewing the AI model's individual-level pricing logic does not satisfy the IRDAI Corporate Governance Regulations' board accountability requirement for product pricing decisions.

India Stack Integration: The Three-Way Compliance Intersection

Bima Sugam's Aadhaar/DigiLocker/UPI integration creates a regulatory intersection with no global equivalent. When an insurer's AI underwriting model uses Aadhaar-fetched data (identity, age, address) and DigiLocker health records as inputs for premium pricing or policy issuance decisions on Bima Sugam:

DPDP Act Section 6 requires that the data be used only for the stated purpose — if the customer consented to Aadhaar data for KYC verification, using it as an input to a risk scoring AI model is a separate purpose requiring separate consent
IRDAI KYC Master Circular requires data accuracy and quality standards for insurance KYC — AI extraction errors from DigiLocker documents that result in policy issuance errors are attributable to the insurer
IRDAI product approval framework governs whether the risk factors being scored by the AI are within the approved product's underwriting criteria

The intersection means that an insurer using Aadhaar/DigiLocker data in AI underwriting on Bima Sugam must obtain DPDP consent for the AI use, maintain IRDAI KYC data quality standards, and confirm the AI's underwriting factors are within approved product criteria — three simultaneous compliance dimensions with no single regulatory framework addressing all three.

Bima Sugam AI Governance Checklist

☐ Product ranking algorithm documentation: ranking criteria documented; conflict-of-interest feature audit completed; commercial signal independence confirmed
☐ Explainability for customer-facing ranking: per-recommendation explanation mechanism in place; explanation grounded in customer needs, not algorithm weights
☐ Chatbot licensing: chatbot operator registered as appropriate intermediary category; suitability assessment AI documented and logged
☐ IRDAI Digital Distribution disclosures in chatbot output: AI involvement, recommendation basis, key terms, human advisor offer — all present at correct customer journey points
☐ IRDAI WG-AI adversarial testing: prompt injection testing for chatbot; adversarial test records maintained; quarterly re-test schedule for customer-facing distribution AI
☐ AI premium pricing actuarial certification: AI pricing output range confirmed within IRDAI-approved rate band; actuary has reviewed AI model's individual-level pricing logic, not only aggregate statistics
☐ India Stack DPDP compliance: separate consent obtained for AI underwriting use of Aadhaar/DigiLocker data; purpose stated as AI underwriting, not only KYC
☐ IRDAI KYC AI quality: AI extraction from DigiLocker documents meets IRDAI KYC data accuracy standards; extraction error rate monitored
☐ MRMF model inventory: all AI models deployed through Bima Sugam (ranking, chatbot recommendation, premium pricing, underwriting) in MRMF model inventory with model cards
☐ IRDAI examination readiness: ranking algorithm documentation, chatbot recommendation logs, pricing AI actuarial certification — packaged for examination