All Issues

AI Security Weekly

Issue #7 — April 2026

The Cyber Insurance Pricing Gap

Published April 27, 2026 11 min read 5 Sections

The Gap, Quantified

01

First Annual Decline in US Cyber Written Premium — The 2024 Inflection

US direct gross written cyber premium fell from $7.25 billion in 2023 to $7.08 billion in 2024 — the first annual decline in the history of the line, drawn from the NAIC Cyber Insurance Supplemental and aggregated in Aon's 2024 US Cyber Market update.1 The American Academy of Actuaries' April 2026 commentary describes this as a market that “nears an inflection point,” framing the data as a structural shift rather than a noise event. The decline arrived alongside a broader pattern: roughly twelve consecutive quarters of negative rate change in standalone cyber, increasing capacity from new entrants, and softening attachment terms across primary and excess layers. None of these forces are mean-reverting on their own. They reinforce each other — capacity competing for share, rates compressing, terms loosening — until something on the loss side breaks the equilibrium. What 2024 records is the first year in which the premium curve declined in absolute dollars while exposure was widely understood to be expanding. This is the structural feature of the current market: the price line is bending in the wrong direction relative to the risk line.

02

Beazley H1 2025: 48.5% Loss Ratio Against a −6.8% Rate Change

In its half-year 2025 results, Beazley — one of the largest cyber underwriters at Lloyd's — reported a 48.5% loss ratio in cyber risks alongside a −6.8% rate change.2 The combination is the headline data point of the soft market: technical profitability is still in the book, but it is being delivered in a pricing environment where year-over-year prices are falling by mid-single-digits. Industry commentary in early 2026 describes this dynamic as a “profitable softening” — carriers continuing to write to a profitable loss ratio while reducing premium dollars per unit of exposure.3 The structural risk is not that the next quarter is unprofitable. The structural risk is that the loss ratio is a lagging indicator of an exposure curve that the rate curve has stopped tracking. When the catch-up arrives — whether through a single tail event, a normalization of frequency, or the materialization of an AI-related loss class — the soft pricing posture is the structural feature that determines how violent the correction becomes.

03

DUAL April 2026: Combined Ratios Could Become Unprofitable by 2027

DUAL's April 2026 cyber market report frames the next twelve to twenty-four months as a binary outcome. Paul Schiavone, EVP, Cyber and Professional Lines, DUAL North America, observes that “our analysis shows that underlying pressures are building. As the market moves towards a more disciplined phase, sustaining long-term capacity and pricing adequacy will be essential not only for insurers, but for the broader relevance of insurance as a mechanism for risk transfer.”3 Ali Khodabakhsh, Head of Cyber, Europe at DUAL, sharpens it: “Two paths now lie ahead. The first leads to gradual price stabilisation over the next twelve months, supporting a sustainable and more resilient market. The second sees existing soft conditions extend into this year and next, increasing the risk of a more severe correction.” The report explicitly warns that combined ratios could approach unprofitable levels in some markets by 2027 if current trajectories hold, and characterizes US conditions as a bellwether for global pricing trends. The message is unusual for its directness from a primary carrier: the market is being told, by the market, that the current pricing posture is not durable.

04

AI Exclusions Are Tightening as Premiums Fall — The Coverage Gap

A January 2026 industry analysis of cyber insurance trends documents the second half of the gap. While premiums are flat or falling, “policy language is tightening in some of the most important areas of emerging risk, especially around artificial intelligence and deepfake-enabled fraud.”4 Carriers are responding to AI exposure by carving the risk out of standard wordings rather than pricing it in. Manuscript exclusions for AI-enabled social engineering, model manipulation, and AI-assisted intrusion are appearing in renewal terms. The buyer-side observation is now widely shared: enterprises deploying AI capability are paying lower cyber premiums in 2026 than in 2024 while receiving narrower AI coverage than they did under the same wordings two years ago. This is the operational shape of the structural mismatch — lower price, narrower coverage, expanding underlying risk surface. The buyer pays less for less. The carrier writes less coverage at less premium. Neither side has confidence in what the AI risk surface costs to underwrite, because the market has not built the shared analytical framework that would tell them.

What the Market Is Saying

Voices From the Carriers, the Regulators, and the Reinsurers

Q1–Q2 2026
Carrier Side DUAL, Beazley, Coalition, Lloyd's LMA
Regulator Side NAIC, NIST, EU Commission
Industry Data Aon, Academy of Actuaries
Time Frame Q1–Q2 2026 disclosures
Posture Bipartisan — carrier & regulator alignment
Direction Toward disciplined underwriting

The 2026 cyber insurance commentary is unusual in its bipartisanship. Carrier-side voices, regulator-side voices, and industry-data voices are converging on the same observation from different starting points. DUAL describes the market as approaching “a more disciplined phase” and characterizes US conditions as the global pricing bellwether.3 Coalition's March 2026 commentary frames the moment as the case for “active insurance” designed to dominate the AI risk landscape rather than retreat from it.5 The Lloyd's Market Association's April 2026 AI Adoption Toolkit signals that the largest specialty market in the world is operationalizing AI underwriting infrastructure as a core competency rather than as a bolt-on.6 Beazley's H1 2025 results disclose the soft-pricing arithmetic in plain numbers.2 The American Academy of Actuaries calls the moment an inflection point.1 The NAIC has moved from advisory posture to operational pilot, and NIST has published the structural assessment of where AI monitoring capability is incomplete.7,8 The European Commission's AI Act is no longer a future event — the high-risk obligations take effect August 2, 2026, in fourteen weeks from this edition's publication.9

Why It Matters

The bipartisan alignment is the signal. When carriers, reinsurers, regulators, and actuaries describe the same conditions in compatible language — soft market, rising exposure, narrowing coverage, accelerating regulatory perimeter — the market has effectively pre-formed the consensus that something needs to change. The disagreement, if there is one, is over the speed of correction and the mechanism by which it arrives. The shape of the answer — a more disciplined underwriting posture for AI-related risk — is no longer contested.

Deep Dive: The Structural Mismatch

$7.25B → $7.08B

US direct gross written cyber premium, 2023–2024 — the first annual decline in the line's history (NAIC Cyber Supplemental, via Aon)1

12+ quarters

Consecutive negative rate change in standalone cyber. DUAL describes US conditions as the global pricing bellwether.3

The structural mismatch in cyber insurance has three layers, each of which is observable in primary disclosures and each of which compounds the others. The first layer is rate. Twelve-plus consecutive quarters of negative rate change is no longer a cycle observation — it is a structural feature of the current market. New entrants and abundant capacity have driven price compression that the loss ratio has not yet caught up to.3 The second layer is exposure. The underlying cyber risk surface has not contracted in proportion to the rate compression. Enterprise AI deployments are extending the cyber attack surface into ML pipelines, model registries, inference infrastructure, and AI-assisted social engineering vectors. The most cited industry quantitative work in early 2026 documents that the AI risk surface is rising on multiple axes simultaneously.10,11 The third layer is coverage. As premiums fall, AI-specific exclusions in cyber wordings are tightening — not loosening. The buyer is paying less for narrower coverage of an expanding risk surface.4 The structural question is not whether the three layers can sustain. They cannot. The structural question is which of them resolves first. If exposure resolves first — through a tail event or through actuarial recognition of the AI risk surface — the result is a sharp rate correction. If coverage resolves first — through standardized AI exclusion language across the market — large segments of AI risk migrate to specialty markets or remain uninsured. If rate resolves first, through reinsurance discipline or regulatory pressure, the soft market ends and pricing adequacy is restored. None of these three resolutions can be characterized as “market continuity.”

The Voices Sounding the Alarm

DUAL — April 2026 cyber market report: combined ratios could approach unprofitable levels in some markets by 2027 if current trajectories hold3
Beazley — H1 2025 results: 48.5% cyber loss ratio against a −6.8% rate change — the soft-pricing arithmetic in plain numbers2
Coalition — March 2026 commentary: the case for “active insurance” designed to dominate the AI risk landscape rather than retreat from it5
Lloyd's LMA — April 2026 AI Adoption Toolkit: Lloyd's market operationalizing AI underwriting infrastructure as core competency6
Academy of Actuaries — April 2026: cyber insurance “nears an inflection point” — the structural framing from the actuarial profession1

The Regulatory Clock Is No Longer Hypothetical

August 2, 2026 — in 14 weeks

EU AI Act high-risk system obligations take effect. Compliance program calendaring is already material for any carrier writing risk in or adjacent to EU AI deployments.

European Commission, AI Act Implementation Timeline

12 states

Participating in the NAIC AI Systems Evaluation Tool multistate pilot through September 2026 — operationalizing examiner capability for AI governance review7

25+ states

Have adopted the NAIC Model Bulletin on the Use of AI Systems by Insurers — over half of US states by April 202612

6 categories

NIST AI 800-4 (March 2026) defines six categories where post-deployment AI monitoring is structurally incomplete: functionality, operational, human factors, security, compliance, large-scale impacts8

May 7–8, 2026

NAIC Insurance Summit AI Forum — the operational convening point for state insurance regulator AI policy in 2026

The regulatory perimeter around AI insurance underwriting tightened decisively in Q1 2026. NIST's Center for AI Standards and Innovation published NIST AI 800-4: Challenges to the Monitoring of Deployed AI Systems in March 2026, organizing post-deployment monitoring into six categories and identifying the specific capability gaps in each.8 The publication is implicitly a regulator-adjacent statement that comprehensive AI risk monitoring is structurally incomplete in 2026 — coverage that assumes mature monitoring is, by NIST's own framing, mispricing the risk. In parallel, the NAIC AI Systems Evaluation Tool moved from advisory document to operational pilot. Twelve participating state insurance departments are now conducting structured AI governance examinations through September 2026, and over half of US states have adopted the NAIC Model Bulletin on the Use of AI Systems by Insurers.7,12 The European Commission's AI Act high-risk obligations take effect on August 2, 2026 — fourteen weeks from this edition's publication. The bipartisan dimension is critical to read correctly. NIST is a US standards body. NAIC is a state-level regulator coordinating body. The European Commission is a multilateral European body. Three different jurisdictions, three different regulatory mechanisms, three different timelines — converging on the same operational expectation: that AI risk underwriting must be conducted against documented, examiner-defensible methodology, not against ad hoc judgment. For US carriers, the NAIC AI Forum on May 7–8, 2026 is the operational convening point where the framework expectations will be discussed in person with state insurance departments. The clock is no longer hypothetical, and the position from which a carrier engages this clock will define its relationship with the regulator for the cycle that follows.

Market Pulse & Reading List

ASI Market Index: 38

Week 17 of 2026, unchanged from Week 16. The Index measures projected annualized AI loss exposure per $1 billion of digital assets at risk — the structural reading of the underlying risk surface, independent of pricing.

ASI Market Index, April 27, 2026

The ASI Market Index sits at 38 this week, stable week-over-week. The Index is calibrated to express projected annualized AI loss exposure as a structural measurement — rising or falling with the underlying risk surface, independent of the soft or hard pricing posture in the market. A composite reading of 38 against the rate-change and premium-decline data above is the quantitative shape of the gap this edition has surfaced. The risk surface measured by the Index is structurally elevated. The price line in the market is structurally suppressed. The gap between the two is the topic of this issue.

01

American Academy of Actuaries — Cyber Insurance Nears an Inflection Point (April 2026)

The actuarial profession's structural framing of the 2024 premium decline. Quantitative anchor for understanding what the NAIC Cyber Supplemental data implies about the next twenty-four months.1

02

DUAL Cyber Market Report — April 2026

The most direct primary-carrier statement of the 2027 unprofitability scenario. The two-paths framing — gradual stabilization versus severe correction — is the operational language the market is now using internally.3

03

Beazley H1 2025 Results — Cyber Risks Performance Review

The 48.5% cyber loss ratio against a −6.8% rate change is the cleanest published expression of the soft-market arithmetic. Direct disclosure from a top-five Lloyd's cyber underwriter.2

04

NIST AI 800-4 — Challenges to the Monitoring of Deployed AI Systems (March 2026)

NIST's structural assessment of what AI monitoring capability cannot yet do reliably. The document defines six categories of post-deployment monitoring and identifies open capability gaps in each — the regulator-adjacent reading of the maturity floor.8

05

EU AI Act Implementation Timeline — August 2, 2026 Effective Date

High-risk AI system obligations under the EU AI Act take effect in fourteen weeks from this edition's publication. For any carrier with EU exposure, the implementation timeline is now a calendaring constraint, not a forward-looking risk.9

Subscribe for Weekly Intelligence

Every Monday. The AI security developments that shape enterprise risk, insurance, and governance — curated by our intelligence team.

Subscribe Free

Sources

American Academy of Actuaries, April 2026 — "Cyber Insurance Nears an Inflection Point" (citing Aon 2024 US Cyber Market Update and NAIC Cyber Insurance Supplemental)

Beazley plc, August 2025 — Half-Year 2025 Results: Cyber Risks Performance Review

Reinsurance News, April 24, 2026 — "Cyber Insurance Market Enters Critical Phase Amid Softening Rates and Rising Exposure: DUAL"

2026 Cyber Insurance Trends Analysis, January 2026 — "Soft Market and AI Exclusions: The 2026 Coverage Trajectory"

Coalition Inc., March 2026 — "Active Insurance Is Built to Dominate the AI Risk Landscape"

Lloyd's Market Association, April 2026 — AI Adoption Toolkit for Managing Agents

National Association of Insurance Commissioners — Big Data and Artificial Intelligence (H) Working Group, AI Systems Evaluation Tool Multistate Pilot

NIST Center for AI Standards and Innovation, March 2026 — "NIST AI 800-4: Challenges to the Monitoring of Deployed AI Systems"

European Commission — EU AI Act Implementation Timeline (Article 6 high-risk system obligations: August 2, 2026 effective date)

Lakera, April 2026 — "Training Data Poisoning: The Complete Guide" (cited as quantitative evidence of expanding AI risk surface)

Zscaler ThreatLabz, April 2026 — "Supply Chain Attacks Surge in March 2026" (cited as expanding AI-adjacent cyber attack surface)

WaterStreet Company, 2026 — "What the NAIC Model Bulletin Means for Insurance AI"