For Insurance Carriers

The Open Standard for AI Insurance Risk

AIRS gives carriers what CVSS cannot: portfolio-level AI risk quantification with direct coverage trigger mapping, built on empirical incident data scored by our intelligence team.

The AI Insurance Readiness Score is a structured, open methodology for evaluating AI-specific insurability risk across model integrity, output liability, supply chain security, regulatory compliance, and systemic resilience. Adopt the standard. Engage our analysts for institutional-grade assessment.

Contact Our Intelligence Team

Domains

Factors

174

Scored Incidents

Supporting Standards

The Underwriting Gap

Why AI Risk Defies Traditional Underwriting

Cyber insurance was built for network intrusions and data breaches — deterministic events with historical actuarial data. AI systems violate every one of these assumptions.

Non-Deterministic Risk

Autonomous agents cascade failures across enterprises in minutes, making loss containment nearly impossible once a compromise is underway. Model poisoning corrupts decision-making at scale — silently and persistently.

No Actuarial Precedent

Prompt injection, RAG poisoning, and Model Context Protocol exploits have no meaningful loss history. Actuaries lack the data required for credible pricing.

Scoring Granularity Gap

Without a purpose-built scoring standard, underwriters cannot differentiate a prompt injection (VSS 53) from an autonomous agent compromise (VSS 81) — yet the coverage implications differ by orders of magnitude.

Multi-Line Trigger

Business Interruption, Cyber Liability, and D&O exposure can trigger simultaneously in an agent compromise scenario. Traditional CVSS compresses both events into a 7–9 range — no actionable signal.

Key Finding

Our incident database reveals that AI-native vulnerability categories — Agent Compromise (avg 80.5), Cascading Agent Failures (avg 80.3), and Memory & Context Poisoning (avg 75.0) — trigger the most severe insurance loss scenarios. Traditional CVSS scores compress these into the 7–9 range, providing no actuarial signal.

Open Methodology, Proprietary Intelligence

What We Publish. What We Protect.

AIRS is an open standard — any carrier can adopt the methodology. Our intelligence infrastructure is what makes it institutional-grade.

The Open Standard

Published methodology that any carrier can adopt for consistent, industry-wide AI risk assessment.

Five assessment domains with defined weights
Twenty-five discrete scoring factors with rubrics
0–100 composite score and four risk tiers
Coverage trigger mapping and claim range estimates
Severity tier classification for underwriting decisions
Integration workflow for existing underwriting platforms

The Intelligence Layer

Seven interlocking proprietary standards that power institutional-grade assessment at enterprise scale.

174 analyst-verified scored incidents (VSS + TSS)
Breach econometrics and loss curve modeling (BEL)
Supply chain dependency mapping and SPoF analysis (SCDM)
Talent migration signals across 200+ companies (TMS)
Regulatory and policy signal tracking (RPS)
Cross-standard correlation and enrichment engines

AIRS Framework

Five Domains. Twenty-Five Factors. One Score.

Each domain evaluates a critical dimension of AI insurance readiness, scored 1–5 per factor with weighted aggregation to a 0–100 composite.

DOMAIN 01

Model Integrity & Validation

Weight: 25%

Training Data Provenance

Adversarial Robustness

Versioning & Rollback

Poisoning Detection

Drift Monitoring

DOMAIN 02

Output Liability & Content Risk

Weight: 20%

Human-in-the-Loop Oversight

Output Validation

Disclosure Practices

Error Correction & Recall

Liability Documentation

DOMAIN 03

AI Supply Chain Security

Weight: 20%

Vendor Due Diligence

Dependency Mapping

Sub-processor Auditing

Contractual Liability

Data Provenance Verification

DOMAIN 04

Regulatory & Compliance Alignment

Weight: 15%

EU AI Act Readiness

U.S. State Law Compliance

Cross-Border Data Transfer

Bias Testing & Fairness

Regulatory Reporting

DOMAIN 05

Systemic Resilience & Continuity

Weight: 20%

Model Diversification

Failure Isolation

Business Continuity

Concentration Monitoring

AI Incident Response

Risk Classification

AIRS Severity Tiers

Four risk tiers translate composite AIRS scores into actionable underwriting decisions — from premium discounts to coverage denial.

Classification	AIRS Range	Underwriting Implication
AI Insurance Ready	80 – 100	Qualifies for affirmative AI coverage with premium discounts of 10–25%. Eligible for broadest coverage terms.
Conditionally Insurable	60 – 79	Coverage available with specific exclusions, sub-limits, and higher premiums. Remediation milestones may unlock broader terms.
Elevated Risk	40 – 59	Limited coverage only, with significant exclusions and mandatory remediation requirements.
Uninsurable	0 – 39	Coverage denied until remediation milestones achieved. Organization provided with roadmap to reach minimum threshold.

Reinsurance Note

Treaty terms can reference AIRS thresholds — for example, requiring cedants to maintain a minimum portfolio-weighted AIRS of 60 to qualify for favorable treaty terms.

Integration

From Scores to Underwriting Decisions

A five-step workflow for carriers to move from raw vulnerability data to actionable underwriting decisions using the AIRS framework.

Score Vulnerabilities with ASI-VSS

Apply the 5-dimension sub-factor rubrics to AI system vulnerabilities in the insured portfolio. Each dimension is scored 0–20; the composite VSS provides a precise, repeatable severity signal.

Map to Coverage Triggers

The Insurance Loss Category sub-factor translates directly to coverage triggers and claim range estimates. Underwriters extract loss category assignments from scored incidents without additional interpretation.

Aggregate by AIRS Domain

Scored incidents aggregate into AI Insurance Readiness Scores across five insurance-relevant domains. Portfolio-level AIRS composites feed directly into underwriting models and pricing engines.

Feed into Underwriting Models

AIRS scores integrate with existing underwriting platforms. Domain scores and composite AIRS provide both granular and summary risk signals compatible with standard pricing architectures.

Monitor and Rescore Quarterly

The AI threat landscape evolves rapidly. Our analysts recommend rescoring high-severity findings quarterly, or immediately following material changes to model version, tool integrations, or deployment architecture.

Carrier Resources

Documentation & Intelligence Products

Resources for carriers evaluating the AIRS framework for portfolio risk management and underwriting integration.

AIRS v1.0 Open Standard Specification

The complete methodology — five domains, twenty-five factors, scoring rubrics, weighted aggregation, and risk tier classification. The foundation for industry-wide AI insurance assessment.

PDF · April 2026 · Open Standard

ASI-VSS Carrier Brief

Purpose-built for underwriting teams. Scoring architecture, D4 insurance loss mapping, severity tier implications, and a five-step carrier integration workflow. Includes empirical data from 174 scored incidents.

PDF · 6 Pages · Institutional Distribution

ASI-VSS Practitioner Brief

For security teams and technical reviewers. Deep dive into the five scoring dimensions, twenty sub-factors, and empirical calibration against real-world AI security incidents.

PDF · 6 Pages · Institutional Distribution

The Open Standard for AI Insurance Risk

Why AI Risk Defies Traditional Underwriting

What We Publish. What We Protect.

The Open Standard

The Intelligence Layer

Five Domains. Twenty-Five Factors. One Score.

AIRS Severity Tiers

From Scores to Underwriting Decisions

Documentation & Intelligence Products

Adopt the Standard. Engage Our Analysts.