What AI governance obligations do IRDAI's 2024 motor insurance regulations create for general insurers?

IRDAI's 2024 regulatory framework for motor insurance — encompassing the IRDAI Master Circular for General Insurance (June 2024, consolidating 13 prior circulars) and the IRDAI (Protection of Policyholders' Interests, Operations and Allied Matters) Regulations 2024 (effective April 1, 2024) — creates three distinct AI governance obligations for India's general insurance sector: (1) Mandatory AI telematics product offering: all general insurers offering motor insurance must provide Pay As You Drive (PAYD) and Pay How You Drive (PHYD) insurance as default first-choice options to motor insurance customers. PAYD adjusts premium based on distance driven, measured through GPS telematics. PHYD adjusts premium based on driving behaviour — speed, braking, acceleration, cornering — measured and scored through AI analysis of IoT device or smartphone sensor data. The board-level AI governance obligation arises because the AI telematics scoring model that converts raw telematics data into a driving behaviour score and a premium adjustment is the core actuarial product: if the AI model is biased, inaccurate, or manipulable, it directly affects policyholders' premiums in discriminatory or unfair ways; (2) Board-approved AI claim settlement policy: the IRDAI Policyholders Regulations 2024 require insurers to have board-approved claim-settlement policies governing AI-assisted claim assessment. For motor claims below ₹50,000, the IRDAI framework permits AI-driven app-based assessment replacing the mandatory surveyor — embedding AI decision-making directly into a regulatory claims obligation. The board-approved policy must cover the AI assessment methodology, human escalation triggers, and policyholder redress mechanisms; (3) End-to-end technology solution governance: IRDAI's telematics product requirements specify end-to-end technology solutions covering onboarding, data collection, pricing, and claim settlement. Each stage involves AI: onboarding AI for KYC and vehicle profiling, data collection AI for telematics signal processing, pricing AI for dynamic premium calculation, and claim assessment AI for damage evaluation. Governing this end-to-end AI pipeline is an implicit obligation within the telematics product offering mandate.

What is the AI telematics scoring governance challenge for PAYD and PHYD motor insurance in India?

Pay As You Drive and Pay How You Drive insurance products collect continuous behavioural data through GPS, accelerometers, gyroscopes, and smartphone sensors — and then apply AI models to convert raw telematics signals into driving behaviour scores and premium adjustments. Five AI governance challenges specific to the Indian telematics context: (1) AI model bias in Indian driving conditions: telematics AI models trained on data from mature market driving environments (UK, US) may systematically misclassify driving behaviour in Indian conditions — urban congestion patterns, emergency braking for pedestrians, narrow road navigation, and monsoon driving all produce telematics signal patterns that differ significantly from the training data. A telematics AI trained on Western highway driving will disproportionately penalise urban Indian drivers for driving patterns that are safe and appropriate in the Indian context; (2) Vehicle type and quality variance: India's 33 crore+ registered vehicles include significant diversity in sensor quality — a 2024 SUV's GPS/accelerometer data is fundamentally different from a 2010 commercial vehicle's. AI scoring models calibrated for high-quality sensor data may produce biased premium adjustments when applied to lower-quality sensor data from older vehicles; (3) Continuous data collection consent: PAYD/PHYD products collect GPS location, speed, and driving behaviour data continuously throughout the policy period — potentially 12 months of detailed location and movement data per policyholder. The DPDP Act 2023 requires purpose-specific consent for personal data processing. The AI telematics system's consent architecture must specify exactly what data is collected, what the AI model does with it, and how the driving behaviour score is calculated — a level of AI transparency that most current telematics consent forms do not provide; (4) Score manipulation and adversarial inputs: PHYD systems that score driving behaviour in real time create adversarial incentives — policyholders may learn to game the scoring system (driving slowly for the first week of the policy, then reverting to normal behaviour; avoiding specific maneuvers the AI penalises regardless of safety). AI telematics models must be robust to such gaming; (5) Insurer AI model selection: IRDAI permits insurers to use their own telematics platforms or partner with third-party telematics service providers. Third-party telematics AI creates the same vendor AI governance challenge as any third-party AI model: the insurer is accountable for the AI scoring outcomes, but the model is opaque unless contractual AI governance provisions require transparency.

How does the DPDP Act 2023 intersect with AI telematics data collection for motor insurance?

AI telematics data collection for motor insurance creates one of the most significant DPDP Act compliance challenges in the insurance sector. Five DPDP Act intersection points: (1) Sensitive personal data classification: GPS location data and continuous movement tracking may be classified as sensitive personal data under DPDP Act 2023 regulations — it reveals where policyholders live, work, worship, seek medical care, and spend time. AI processing of sensitive personal data carries enhanced consent and purpose-limitation obligations; (2) Purpose-specific consent for AI scoring: the DPDP Act requires that consent be specific to the purpose for which data is processed. A general consent to 'telematics data collection for insurance purposes' is insufficient — the consent must specify that the GPS data will be processed by an AI model to generate a driving behaviour score that adjusts the motor insurance premium. If the insurer also uses telematics data for other purposes (marketing, product development, sharing with reinsurers), each purpose requires separate consent; (3) Data minimisation for AI models: DPDP Act requires data minimisation — collecting only the data necessary for the stated purpose. A PHYD AI model that requires continuous high-frequency GPS data (position every second) when a lower-frequency sample (every 30 seconds) would produce equivalent scoring accuracy is not data-minimised. Insurers must assess whether their AI telematics models use more data than necessary for accurate scoring; (4) Automated decision-making and right of explanation: when a PAYD/PHYD AI model produces a premium adjustment or claim assessment decision, the DPDP Act's framework for automated decision-making requires that policyholders have access to the basis of AI decisions that affect them. A policyholder who receives a premium increase because of low driving behaviour scores must be able to understand why the AI scored them poorly; (5) Data retention limitations: DPDP Act requires that personal data not be retained beyond the purpose for which it was collected. For telematics data, this creates a question: how long should an insurer retain GPS and driving behaviour data after a policy expires? If the AI model uses historical telematics data for pricing renewal policies, the retention justification must be documented and disclosed.

What board-level AI governance obligations does the IRDAI Policyholders Regulations 2024 create for AI-driven motor claims assessment?

The IRDAI (Protection of Policyholders' Interests, Operations and Allied Matters) Regulations 2024 create a board-level AI governance obligation through the requirement for a board-approved claim-settlement policy that governs AI-assisted claim assessment. Four board-level AI governance dimensions: (1) Board approval of AI assessment methodology: for AI-driven app-based motor claim assessment below ₹50,000, the board-approved claim-settlement policy must specify the AI methodology used — what the AI model assesses (vehicle damage photos, repair estimates, parts costs), how it reaches a claim settlement figure, and what confidence thresholds trigger human review versus automated settlement; (2) Human escalation criteria in board policy: the board policy must define the criteria under which an AI claim assessment is escalated to human review — claims above a threshold, AI confidence below a threshold, policyholders who contest the AI assessment. IRDAI's policyholders protection requirements mean the AI assessment cannot be the final word in all cases; (3) Accountability for AI claim outcomes: when an AI claim assessment produces an incorrect outcome — a damaged vehicle assessed as a total loss when it is repairable, or a genuine total loss assessed as repairable — the board-approved policy must specify the accountability chain. The insurer remains accountable for AI claim assessment accuracy regardless of whether the AI was developed in-house or procured from a third-party motor claim assessment vendor; (4) Policyholder redress for AI decisions: IRDAI's policyholders protection framework requires accessible redress for policyholders who dispute claim settlement decisions. When the disputed decision was made by an AI model, the redress process must include a human review of the AI assessment — not merely a reconfirmation of the AI's output. The board-approved policy must specify the AI dispute resolution process, including the human review mechanism and timelines.

AI Governance · IRDAI · Motor Insurance · AI Telematics · PAYD · PHYD · India BFSI

IRDAI Motor Insurance AI Telematics Governance 2024: Mandatory Pay As You Drive AI Pricing, Board-Approved AI Claim Settlement Policies, AI App Claims and DPDP Continuous Telematics Data Obligations for 33 General Insurers India

IRDAI Master Circular for General Insurance (June 2024) and IRDAI (Protection of Policyholders' Interests) Regulations 2024 (effective April 1, 2024) mandate AI telematics governance across India's motor insurance sector. All 33 SEBI-registered general insurers must offer Pay As You Drive (PAYD) and Pay How You Drive (PHYD) as default first-choice options — embedding AI telematics scoring models into the core insurance product. Board-approved policies must govern AI-assisted claim settlement. For motor claims below ₹50,000, AI app-based assessment replaces the mandatory surveyor — making AI decision-making a regulatory entitlement, not just a cost-saving choice. Continuous GPS and driving behaviour data collection creates DPDP Act 2023 consent and purpose-limitation obligations. India has 33+ crore registered vehicles. Zero global competitor coverage.

Zytra Research Team · April 2024 · 12 min read

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

India's motor insurance sector is being reshaped by a regulatory mandate that is simultaneously an AI product requirement. IRDAI's 2024 requirements don't just permit AI telematics — they require general insurers to offer PAYD and PHYD as default options. Every general insurer that sells motor insurance must now operate AI systems that continuously collect driving behaviour data, score it against an actuarial model, and adjust premiums accordingly. This is AI governance at scale: 33 general insurers, 33 crore registered vehicles.

The AI governance challenge is that the AI scoring model is now the insurance product. A biased telematics model doesn't just produce inaccurate predictions — it produces unfair premiums. A policyholder in urban congestion who brakes frequently because of traffic is not a reckless driver, but a PHYD model trained on highway driving will penalise that braking pattern. The board-approved claim-settlement policy requirement is IRDAI's mechanism for requiring boards to own this AI risk — not delegate it entirely to actuarial teams and technology vendors.

The Rs 50,000 AI Claims Threshold

IRDAI's permission for AI app-based claim assessment below Rs 50,000 without a surveyor creates a significant AI governance moment. Motor surveyor assessment has been a mandatory consumer protection mechanism in India's insurance framework. Replacing it with AI assessment for the majority of small motor claims (most hail damage, minor accident damage, and windscreen claims fall below Rs 50,000) means AI accuracy directly determines claim fairness for retail policyholders at scale. The board-approval requirement for this AI methodology is proportionate to its consumer impact.

IRDAI Motor Insurance AI Telematics Governance Checklist

☐ PAYD/PHYD product offered: Pay As You Drive and Pay How You Drive options available as default first-choice motor products; AI telematics scoring model operational; end-to-end technology solution (onboarding, data collection, pricing, claims) deployed
☐ AI telematics model governance: board-level awareness of AI telematics scoring methodology; actuarial validation of AI scoring model for Indian driving conditions; bias assessment for urban vs rural driving patterns, vehicle age/type variance
☐ Board-approved claim settlement AI policy: board-approved policy for AI-assisted claim assessment; AI methodology for sub-Rs 50,000 claims specified; human escalation criteria defined; AI dispute resolution process documented
☐ DPDP Act telematics consent: purpose-specific consent for AI processing of GPS and driving behaviour data; consent covers AI scoring model operation and premium adjustment; data minimisation assessment for AI model inputs; retention limitation policy for post-policy telematics data
☐ Policyholder right of explanation: process for policyholders to receive basis of AI driving score and premium adjustment; process for policyholders to contest AI claim settlement decisions with human review
☐ Telematics vendor AI governance: due diligence on third-party telematics AI providers; contractual AI transparency, audit rights, and data handling obligations
☐ IRDAI examination readiness: telematics product documentation, board-approved claims AI policy, consent architecture, actuary validation of AI model — packaged for IRDAI examination