LO
ConstraintCognitive automation2024-07-11
New York's insurance regulator expects insurers to test AI underwriting for unfair discrimination before use and regularly after, and to tell applicants why an AI process turned them away
Insurance underwriteroccupation page →
Event date / reported
2024-07-11
Evidence stage
ConstraintFailure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.
Tasks this bears on
Underwriting standard applications
Accepting or declining straightforward applications — healthy applicants, small sums, standard small-business risks — against the insurer's rules.
Automating✓ Evidence-backed
Maintaining the rules the machine decides by
Writing and tuning automated underwriting rules, checking their decisions, and watching for unfair or drifting outcomes.
New task✓ Evidence-backed
Where this applies
New York State only: a circular letter from the Department of Financial Services to every insurer authorised in the state, issued in final form on 11 July 2024 after a proposed version in January. Read for its verbs. The letter uses must only narrowly — insurers must be able to demonstrate that the external consumer data they use is not prohibited, and must take and keep records of complaints about AI systems and external data. What bears on underwriting work is written as supervisory expectation: an insurer should not use AI systems or external consumer data in underwriting or pricing unless it can establish they are not unfairly or unlawfully discriminatory; that testing should be done before a system is put into production, on a regular cadence and whenever the model or data changes materially, with testing of model output, including drift, at least annually, documented and available to the Department; insurers should have competent and qualified personnel to execute and oversee this risk management, and may not rely solely on a vendor's claim of non-discrimination. When an AI-based process cannot underwrite an applicant, the insurer should tell them why in writing within 15 days and continue underwriting them through the non-AI process. It does not say who inside an insurer does the testing — underwriters, actuaries or a model-risk team — and it does not require a person to decide any individual application.
What this means
Where automated underwriting is used in New York, the regulator expects someone to test the machine's decisions for unfair discrimination before it goes live and to keep testing it, including for drift, with qualified staff answerable for it. That is the new work the underwriter page describes — checking the rules the machine decides by — written down as a supervisory expectation.
What it does not yet show
It is guidance written mostly as should, for one state; it does not say whether underwriters, actuaries or a separate model-risk team do the testing, and it says nothing about how many underwriters there are.
What you can check
Open the Department of Financial Services' Insurance Circular Letter No. 7 (2024) and find the sentence beginning "Unfair or unlawful discrimination testing and analysis should be administered prior to putting AIS into production".
Does it change the assessment?
No. The impact index is never moved by a single event. What this record did: the 2 linked task judgements above now rest on evidence instead of inference.
Source
New York State Department of Financial Services — Insurance Circular Letter No. 7 (2024): Use of Artificial Intelligence Systems and External Consumer Data and Information Sources in Insurance Underwriting and Pricing (11 July 2024) · verified 2026-09-27 · Claude (VOLO agent) · interpreted 2026-09-27 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.