Accelerated Underwriting

Accelerated underwriting is a life insurance underwriting path that clears an applicant without the paramedical exam and the blood or urine collection, putting outside data and a model where that evidence would have been. The NAIC working group that studied accelerated underwriting from 2019 defined it as the use of big data, artificial intelligence, and machine learning to underwrite life insurance in an expedited manner. Its wording is hedged: the process is typically used to replace all or part of traditional underwriting, letting some applicants have certain medical requirements waived.

What an accelerated underwriting program replaces

Traditional life underwriting assesses physical health alongside financial and behavioral evidence, then places the applicant in a risk class. An accelerated program keeps the risk classes and changes where the evidence comes from. The NAIC’s examples of inputs run to customer disclosures, prescription history, digital health records, credit attributes, medical information bureau data, public records, motor vehicle reports, smartphone and wearable data, purchasing history, and social media. A predictive model or machine learning algorithm then does part of what an underwriter would have done.

How does an application move through an accelerated underwriting program?

Whether an application takes the accelerated path is decided from data. The NAIC’s topic page says insurers increasingly rely on outside data, predictive models, and algorithmic or machine learning techniques to evaluate risk and decide who is eligible for an accelerated pathway. Presentations to the working group indicated two main uses: triage, where applicants who do not clear are routed back into traditional underwriting while the rest continue on the accelerated path, and direct assignment of risk classes.

Accelerated underwriting alone, the same page cautions, does not always lead to a policy: for some applicants the available data will be too thin to judge the risk, and they still have to complete the traditional process, physical exam included.

Where a person comes in is not uniform. The working group’s 2022 report records presentations indicating that adverse underwriting decisions are sometimes, not always, reviewed by human underwriters, and states more generally that the exact parameters of a program vary by insurer.

Two things it gets confused with

It is not simplified issue, which the NAIC calls simplified underwriting. That route also drops the exam and the fluids, but the NAIC describes the trade as made in exchange for generally higher premiums. An accelerated program goes after the missing evidence itself.

It is also not a synonym for full automation, in either direction. An appendix to the working group’s report, borrowing a definition drafted for a survey of private passenger auto insurers, separates automated approval or denial, taken with no human involvement on that file, from a model that only feeds a decision the underwriter still owns.

Why regulators keep coming back to it

Medical evidence has, in the NAIC’s phrase, a scientific linkage with mortality. The non-medical behavioral data these programs can lean on does not carry that link automatically: gym membership, occupation, marital status, family size, grocery habits, wearables, and credit attributes may, the working group wrote, lead to questionable conclusions without a reasonable explanation. That is the setup for proxy discrimination: a facially neutral variable that tracks a protected class closely enough to move outcomes.

Transparency is the other half. The same report warns that non-traditional data may not carry the protections the Fair Credit Reporting Act attaches to consumer reports: there may be no clear path for a consumer to learn how a data element affected the application, and the type and purpose of the data accessed need not be disclosed. An applicant can be scored on records they never knew were pulled.

The rules that actually reach it

The working group’s position, stated in its 2022 report and again in the referral adopted with its later guidance, was that existing regulations apply to accelerated underwriting programs in the same way as to traditional ones.

The NAIC’s Life Insurance and Annuities (A) Committee adopted Accelerated Underwriting in Life Insurance Regulatory Guidance and Considerations on August 14, 2024. It is written for insurance departments rather than carriers, as a framework to reference when a department reviews a life insurer’s program, and the working group contemplated its use in market conduct reviews. It runs not on the educational report’s definition but on the reporting definition of accelerated underwriting added to the Life and Annuity Market Conduct Annual Statement in 2022, which catches any use of AI or machine learning drawing on non-medical third-party data; the working group states it takes the two to be consistent.

The guidance has three parts: regulatory considerations, strategies for review, and requests for information. The third lists eleven sample questions, offered as examples of what a department may want to submit to a life insurer. Five concern the external data and its disclosure: which information a program uses, how the company obtains it, what it discloses to applicants, and how the data is stored after underwriting and ultimately destroyed. Two ask what recourse an applicant has, after an adverse underwriting decision based on that data or to correct a mistake in it. Three turn to how the program is checked: how data sets, predictive models, and machine learning algorithms are validated, tested, or audited, including for potential adverse consumer outcomes, whether that work is done internally or by a third party, and how the company ensures its models rest on sound actuarial principles. The last asks how a company ensures that external data’s correlation to risk is not outweighed by its correlation to a protected class.

New York’s Circular Letter No. 7 reaches AI systems and external data in underwriting and pricing, and it lands awkwardly here: its own definition of external consumer data leaves out several inputs these programs lean on, motor vehicle reports and prescription drug data among them. Which of its two assessments then applies belongs to the Circular Letter guide.

Colorado began its rulemaking under SB 21-169 with life insurance underwriting; that entry carries dates and scope.

What a life insurer should keep on record for such a program, and where these programs predictably fail, is the subject of AI in life insurance underwriting.