KYE Insurance Underwriting & Claims Authority Pack™ — defensible, fair, auditable AI underwriting & claims decisions.
AI now does risk scoring, underwriting, and claims adjudication — and regulators demand the decision be fair, explainable, and contestable when challenged. KYE Protocol™ governs the authority and evidence of AI-assisted insurance decisions and proves they are fair and contestable: which named underwriter or adjuster authorised the adverse decision, the recorded adverse-action reason-code and the evidence behind it, the proxy-discrimination fairness-evidence captured before it proceeds, and a signed, replay-provable Evidence Pack™ per decision — with an appeal / contestability record so any decision can be reconstructed and contested. KYE Protocol™ governs whether the AI-assisted decision may proceed and proves it is fair — it does not price the risk, write the policy, set the risk appetite, or replace the actuary, underwriter, or adjuster.
AI now scores the risk and adjudicates the claim — and the decline, the adverse rate, and the claim denial are the moments accountability concentrates.
Generative and predictive underwriting models, automated risk scorers, and AI claims-adjudication tools are producing decisions that move quickly toward an applicant’s decline, an adverse rate, or a claimant’s denial. The high-value problem is not the modelling — it is the action boundary and its fairness and defensibility. Four facts converge:
- The consequential moment is the adverse decision — not the score. A model’s score is inert; a declined application, an adverse-rate offer, an applied exclusion, or a denied claim is consequential. When the decision is challenged — a market-conduct exam, an applicant’s adverse-action appeal, a GDPR Art. 22 contest — the regulator demands to see who authorised it, why, and that it was tested for unfair discrimination.
- Anti-discrimination is now a hard requirement. NAIC, Colorado SB21-169, and the NYDFS circular require insurers to test AI and external-data-driven decisions for unfair discrimination and document it. KYE Protocol™ refuses any consequential decision whose proxy-discrimination fairness-evidence was not captured — an untested model output never proceeds.
- Adverse decisions must be explainable. Colorado, the NYDFS circular, and GDPR Art. 22 require a meaningful, specific reason for an adverse decision. KYE Protocol™ refuses a black-box decline: no consequential adverse action proceeds without a recorded adverse-action reason-code and the evidence behind it.
- This is a governance wedge, not a pricing engine. KYE Protocol™ does not compete with the actuarial model, the rating engine, or the claims platform. It governs the action boundary they feed — the named-authority + adverse-action explainability + fairness-evidence + Evidence Pack™ + appeal layer the AI insurance ecosystem currently lacks.
Survives a market-conduct exam, an adverse-action appeal, or an Art. 22 contest — fairness-tested, reason-recorded, and derivable from public keys alone.
- No black-box declines, by construction. Every adverse underwriting or claims decision that proceeds toward a consequential action must carry a recorded adverse-action reason-code and the evidence behind it. A consequential adverse action without a recorded reason is refused at the action-admissibility gate and never proceeds.
- Adverse decisions are authority-bound. Every decline, adverse rate, exclusion, denial, or rescission maps to a recorded named-authority decision — the agent, the decision, the action, and the named underwriter or adjuster under whose authority it proceeds. An AI authorised for one purpose cannot proceed under another.
- Fairness-evidence on every decision. Every consequential decision carries captured proxy-discrimination fairness-evidence — the protected-class proxy test, the disparate-impact result, the data fields tested — so it is provably tested against unfair discrimination under the NAIC, Colorado, and NYDFS regimes.
- Replay-provable Evidence Pack™. Every decision emits a signed Evidence Pack™ binding the authority, the adverse-action reason, the fairness-evidence, and the decision basis — reconstructable and valid at T=0, derivable from published keys alone, retained under WORM — the defensibility artefact a regulator or an applicant can verify offline.
- Contestable when challenged. Every decision carries an appeal / contestability record so an Art. 22 request for human intervention, a market-conduct inquiry, or an adverse-action appeal can reconstruct it exactly as made and contest it through a recorded route. Bound to the NAIC AI Bulletin, Colorado SB21-169, the EU AI Act Annex III insurance use-cases, the NYDFS AI circular, and GDPR Art. 22 — each with a 90-day attestation cadence.
Every consequential insurance decision — authority-bound, fairness-tested, and evidenced at the action boundary.
One coherent spine governs three specializations — underwriting, claims, and fraud-review — with no parallel packs. Each AI-assisted decision that moves toward a consequential adverse action flows through the same five rules, on the canonical KYE Protocol™ envelopes.
- 1 — Decision proposed. An AI model produces an underwriting decision (decline, adverse rate, exclusion) or a claims determination (denial, reduction, fraud-flag) that begins to move toward being applied to a real applicant or claimant.
- 2 — Authority + reason check. The Action Admissibility™ Gate verifies the named-authority under which the decision proceeds and that a meaningful adverse-action reason-code and its evidence are recorded, under the §25 Edge Governance Safety Floor. No authority, or a black-box decline = no action.
- 3 — Fairness-evidence captured. Every consequential decision carries its proxy-discrimination fairness-evidence — the protected-class proxy test, the disparate-impact result, the data fields tested — sealed as a KYE Fairness Evidence Pack™ before it proceeds.
- 4 — Evidence Pack™ + appeal record sealed. The runtime emits kye.purpose.request.v1 + kye.purpose.admissibility.v1 + kye.evidence.decision_map.v1 + kye.evidence.pack.v1 + kye.replay.context_seal.v1 in lockstep, binding the authority, the adverse-action reason, the fairness-evidence, and an appeal / contestability record into a signed, replay-provable, WORM-retained Evidence Pack™ — reconstructable for a regulator or an applicant when the decision is challenged.
Bound to the insurance AI-governance, anti-discrimination, and automated-decision perimeter.
The pack binds the canonical KYE™ artefact set to the insurance underwriting & claims AI perimeter. Every claim resolves to a control row on the bound framework — the five regimes are consumed by the rule pack, never re-mapped (honest scope: KYE™ maps only the authority / evidence / fairness / defensibility slices, and cedes the actuarial pricing / risk appetite / model design to the carrier and the actuary).
| Framework | Control area | Pack coverage |
|---|---|---|
| NAIC Model Bulletin on the Use of AI by Insurers | Named accountability on the AI decision, adverse-action documentation, unfair-discrimination testing evidence | partial |
| Colorado SB21-169 | Adverse-action reason explainability, external-data proxy-discrimination evidence, named-authority | partial |
| EU AI Act — Annex III high-risk insurance | Human-oversight authority (Art. 14), record-keeping / logging (Art. 12), transparency & contestability | partial |
| NYDFS Insurance Circular Letter on AI | Senior-management accountability, unfair-discrimination testing, consumer transparency & appeal record | partial |
| GDPR Article 22 | Human-involvement safeguard, meaningful information about the logic, right to contest & human intervention | partial |
Honest scope. KYE Protocol™ governs the authority, adverse-action reason, fairness-evidence, evidence, and contestability of the AI-assisted decision at the action boundary — whether the decision may proceed and how it came into existence, so it is defensible and fair when challenged. It does not price the risk, write the policy, set the risk appetite, render the actuarial judgment, or replace the actuary, underwriter, or adjuster. Partial coverage means the bound surface satisfies the authority / evidence / fairness / defensibility slice of the control area when paired with the carrier’s own actuarial and model-governance judgment. KYE™ complements the actuarial model, the rating engine, and the claims platform — it does not compete with them.
Qualified insurance-AI partners — apply through the Foundry.
The KYE Insurance Underwriting & Claims Authority Pack™ is a §68 sector product productised through the KYE Sector Pack Foundry™ Build tier, with Starter, Enterprise, and Regulated commercial tiers; commercial distribution is value-based, qualification-gated, and disclosed under NDA to qualified applicants.