By James W. Moore
Editor’s Note
Last week, we looked at how insurers are beginning to identify, exclude, price, and cover AI-related liability. This week, one of the companies helping drive the technology offered a striking assessment of how difficult that task could become.
In its October 26 S-1 filing, Anthropic warned that a catastrophic AI incident could produce losses beyond the capacity or underwriting expertise of any single insurer, potentially even the industry collectively. That is a risk disclosure, not a prediction, and it should not be mistaken for a claim that one event would exhaust the insurance industry’s reserves. But it is remarkable to see an AI developer raise the question of insurance market capacity in a securities filing.
Meanwhile, the insurance industry continued building, buying, and deploying AI. Guidewire announced a multibillion-dollar acquisition aimed at agentic underwriting. California expanded its approach to long-term insurer solvency to include AI. BlackRock found a substantial gap between how executives and other insurance professionals assess AI adoption.
The technology is moving into the industry’s operating infrastructure, while insurers and regulators are still working through what that means for risk, authority, and financial resilience.
James W. Moore
Editor-in-Chief, InsuranceIndustry.ai
BlackRock Finds a 22-Point Divide in Insurance AI Adoption
BlackRock’s 2026 Global Insurance Report found that 73% of C-suite respondents reported having at least one AI use case operating at scale, compared with 51% of non-C-suite respondents.
The 22-percentage-point difference stands out in a survey of 463 insurance professionals across 31 countries. Across the full sample, 60% reported at least one scaled or embedded AI use case. Underwriting, claims, and compliance were among the areas where adoption was most advanced.
There are several possible explanations for the gap. Executives may have greater visibility into enterprise-wide projects, while operational leaders may judge progress by whether technology is fully integrated into everyday work. Respondents may also differ in what they consider a scaled deployment.
The survey does not establish which explanation is correct. It does establish that different groups of insurance professionals are reporting materially different levels of AI progress.
Why it matters: Access to AI, implementation of AI, and meaningful operational adoption are not interchangeable measures. An insurer may have numerous successful deployments without those systems changing how most of its people work.
For boards and senior management, the useful question may be less about how many AI projects have reached production and more about how much of the organization’s actual work those projects now perform.
Guidewire Makes a $2.1 Billion Bet on Agentic Underwriting
Guidewire announced an agreement on October 27 to acquire London-based hyperexponential for $1.7 billion in cash, with up to another $400 million in contingent consideration tied to recurring revenue targets.
Hyperexponential develops underwriting and pricing technology for commercial and specialty P&C insurance. According to Guidewire, the company’s platform has supported underwriting decisions involving more than $75 billion in annual premium across insurers, reinsurers, and MGAs.
Guidewire describes the combination as a foundation for agentic underwriting, where AI systems work alongside underwriters and actuaries to evaluate risks, develop prices, and support decisions.
The acquisition remains subject to closing conditions, and its commercial results are still to be demonstrated. But the purchase price is a substantial signal about the strategic importance Guidewire places on underwriting intelligence.
Why it matters: Much of the early discussion around agentic insurance technology has focused on startups building automated workflows around established core systems. Guidewire’s proposed acquisition points toward another possibility: the core technology providers may incorporate those capabilities directly.
That does not eliminate opportunities for specialist vendors. It does change the competitive environment in which they will operate.
The longer-term question is whether underwriting intelligence remains a separate application or becomes a native capability of the platforms insurers already depend on.
California Adds AI to Long-Term Insurer Solvency Oversight
California Insurance Commissioner Ricardo Lara finalized a new Long-Term Solvency Planning Regulation on October 27 that explicitly includes artificial intelligence among the emerging risks insurers must consider in assessing their future financial condition.
The regulation applies to California-domiciled insurers and establishes enhanced requirements for companies writing more than $50 million in nationwide direct annual premiums.
Alongside climate change, cybersecurity, and technological disruption, insurers must evaluate how AI-related developments could affect their financial strength, operations, investments, and underwriting strategies over time.
This is not a new model-testing standard or a separate set of detailed AI governance controls. AI is one component of a broader, forward-looking solvency framework.
Why it matters: Much of the insurance regulatory discussion around AI has concentrated on individual decisions: underwriting fairness, consumer outcomes, documentation, explainability, and model oversight.
California’s action places AI within a wider institutional question. What could technological change mean for an insurer’s long-term ability to remain financially sound and meet its obligations?
That includes risks arising from an insurer’s own use of technology, but also the consequences of AI-driven changes to the businesses and exposures it insures.
AI governance and enterprise financial risk management are becoming increasingly difficult to separate.
UK Regulators Find AI Agents Widespread, but Broad Autonomy Rare
The Bank of England and Financial Conduct Authority published their 2026 survey of artificial intelligence in UK financial services on October 26.
Across 196 responding firms, including insurers, 86% reported using AI, up from 75% in 2024. Eighty percent reported having AI agents deployed, with another 14% planning deployment during the following year.
The distinction between deployment and authority is particularly revealing.
Eighty percent reported agents in production providing decision support, where humans make and execute the decisions. Only 2% reported agents with broad autonomy.
Third-party dependency is also becoming more significant. Third-party implementations represented 39% of reported AI systems and 53% of systems classified as highly material.
The survey also identified a gap between initial governance procedures and some ongoing monitoring practices.
Why it matters: Agent deployment does not necessarily mean decision-making authority has been delegated.
A system that recommends a claim outcome, an underwriting decision, or a financial transaction presents a different control problem from one authorized to execute the action.
At the same time, human approval alone does not guarantee effective oversight. The organization still needs to understand what the system recommends, what evidence it uses, and how its behavior changes after deployment.
The UK findings offer an unusually useful view of where financial institutions are actually placing authority, rather than where the agentic AI marketing narrative suggests they are heading.
Verisk Brings Insurance Data Into Copilot, Claude, and Amazon AI
Verisk announced October 28 that it is expanding access to its proprietary insurance intelligence through Microsoft Copilot, Amazon Quick, and Anthropic’s Claude.
The integrations bring insurance-specific information, including underwriting, premium audit, and property-weather intelligence, into general-purpose enterprise AI environments.
The development reflects an important change in the relationship between insurance data providers and the software through which that information is used.
Traditionally, insurers accessed specialized datasets and analytical tools through dedicated applications or integrations. General-purpose AI interfaces create another route, allowing authorized users to work with industry-specific information without necessarily moving between multiple standalone systems.
The value of the announcement will depend on implementation, licensing, reliability, and how effectively insurers incorporate the capabilities into existing workflows.
Why it matters: The strategic value of proprietary insurance data may increasingly depend on how easily that data can be used by AI systems operating across an enterprise.
That creates opportunities for established information providers such as Verisk, but it also changes the competitive position of software products whose main function is presenting or organizing that information.
The underlying data and its permitted uses remain important. The interface through which people access it may become less fixed.
Kin Opens Home Insurance Quoting to AI Shopping Agents
Home insurer Kin announced October 23 that it has enabled WebMCP within its digital quoting experience, allowing AI shopping systems to interact more directly with its online insurance process.
Kin describes itself as the first U.S. home insurer to offer the capability, although that first-mover designation is the company’s own claim.
The development goes beyond using a chatbot to answer questions about insurance products. It is an effort to make the insurer’s shopping interface usable by AI agents acting on behalf of consumers.
That distinction matters for insurers and distributors whose customer acquisition strategies depend heavily on attracting people to websites, comparison platforms, and digital quoting forms.
Why it matters: If consumers increasingly delegate parts of insurance shopping to AI assistants, the systems influencing carrier discovery and quote comparisons may change.
The insurer’s website still matters, but its ability to communicate with automated shopping tools could become another dimension of distribution.
It is too early to conclude that AI-mediated shopping will materially displace existing channels. Kin’s announcement nevertheless offers a concrete example of a carrier preparing for that possibility rather than waiting for customer behavior to change.
PolicyMe Reports 59% Faster Life Underwriting Case Turnaround
Canadian insurance distributor PolicyMe has introduced an internally developed AI underwriting copilot intended to help human underwriters review complex life insurance applications.
According to the company, the system reduced turnaround time for the affected cases by 59%, based on internal measurements.
The copilot brings medical records, laboratory results, and supporting documents into a consolidated workspace, produces structured case summaries, recommends decisions with supporting evidence, and prepares draft explanations for certain underwriting outcomes.
Human underwriters retain final decision authority.
The reported improvement should be interpreted narrowly. It concerns the cases measured by PolicyMe, not necessarily the complete application-to-policy process, and the results have not been independently validated.
Why it matters: Some of the most practical AI gains in underwriting may come from reducing the time professionals spend assembling and interpreting evidence rather than replacing their judgment.
Life insurance is particularly well suited to examining this distinction because complex applications can involve substantial medical documentation and consequential decisions.
PolicyMe’s reported results are promising, but the larger test is whether efficiency improvements can be sustained while preserving underwriting quality, documentation, and accountability.
Sources
- BlackRock, 2026 Global Insurance Report, October 2026.
- Guidewire, Agreement to Acquire hyperexponential, October 27, 2026.
- California Department of Insurance, Long-Term Solvency Planning Regulation, October 27, 2026.
- Bank of England and FCA, Artificial Intelligence in UK Financial Services, October 26, 2026.
- Verisk, Expansion of Insurance Intelligence Across Enterprise AI Platforms, October 28, 2026.
- Kin, WebMCP for AI Shopping, October 23, 2026.
- Coverager, PolicyMe Launches AI Underwriting Copilot, October 2026.
- Anthropic, Form S-1 Registration Statement, October 26, 2026.
More AI Insights
Catch up on previous editions of AI Insights, our weekly review of the developments shaping AI and insurance.
Browse AI Insights Archive
AI Disclaimer: This content was created with assistance from artificial intelligence technology. While content is based on factual information from the source material, readers should verify all details directly with the respective sources before making business decisions.
