Reading scan time: 5 minutes
Listening time: 8 mins

Your Friday Five

This week we're doing something slightly different. One topic, "Silent A.I." Three articles. All on the single most consequential coverage development facing financial institutions in 2026.

Most boards haven't been fully briefed on it until now!

• AI-related lawsuits surged 978% from 2021 to 2025. Most of the policies covering those defendants never mention the word "AI."
• Silent AI coverage is disappearing from your program right now, not through a single mandate, but renewal by renewal, endorsement by endorsement.
• In the last 12 months, at least six purpose-built AI insurance products hit the market. Three major carriers filed to exclude AI risk entirely. Both moves are happening at the same time.

Three aspects of Silent AI that we think deserve your attention…First, If you'd rather listen check out the audio version.

Silent AI — What It Is, Why It's Not Silent Cyber, and Why Your Board Needs the Vocabulary Now

Summary

The insurance industry has a name for the coverage problem most boards haven't heard of yet.

“Silent AI,” refers to AI-driven risks neither explicitly included nor excluded in existing insurance policies. Most policies your institution carries today were written before AI became embedded in business operations. They don't contemplate AI-generated losses and they don't define AI-related acts.

That ambiguity cuts both ways, creating unintended liability for carriers and unexpected gaps for policyholders.

Three mechanisms drive the exposure. Embedded AI is the AI your institution knowingly deploys in credit underwriting, fraud detection, claims processing, submissions triage, or pricing. Self-procured AI is the AI your employees use independently without institutional oversight. And third-party AI is the AI your vendors deploy on your behalf, inside your data, without explicit governance. That third category is where most institutions have the least visibility.

Every institution has all three. The question is which one your risk committee can see and which two it can't.

(sources: Kennedys Law, Swiss Re SONAR 2024, DAC Beachcroft)

So what?

Silent cyber was a discrete new peril.
​
Lloyd's mandated clarity, carriers excluded it, and the market rebuilt affirmative coverage on top. Silent AI doesn't work that way. As DAC Beachcroft's Jonathan Hopkins put it: “AI often amplifies traditional risks rather than introducing entirely new exposure. A carrier can't exclude AI from a D&O policy without gutting coverage for board decisions that increasingly rely on AI-generated data.”

There will be no single mandate to fix this. It will be resolved line by line, carrier by carrier, renewal by renewal. As long as policies remain silent on AI, the legal principle that ambiguous language is interpreted in the policyholder's favor works in your direction. As carriers introduce explicit endorsements, that window narrows.

Monday morning action: Ask your risk manager two questions: What AI are we deploying intentionally? What AI are our employees using without our knowledge? If the second answer is "we don't know," that's the exposure.

The Coverage Cliff — How Silent AI Hits Six Lines at Once

Summary

The market put a number on it this week.

A major reinsurance broker's white paper, published this week in partnership with MIT and Testudo, tracked over 700 cumulative GenAI-related lawsuits in the United States between 2020 and 2025, a 978% surge since 2021, with acceleration jumping from 59% in 2023–2024 to 137% in 2024–2025. Most headline volume involves tech developers and AI platforms. But the litigation that matters most to FI buyers sits in adjacent categories: algorithmic discrimination triggering EPLI, AI-washing securities actions triggering D&O (53 AI-related securities class actions filed through H1 2025), and AI-generated advisory errors creating E&O boundary disputes still unresolved in court. A global survey of 1,250 companies found 57% flagged AI errors and hallucinations as their top risk, yet traditional insurance responds only in fragments.

The disconnect between litigation velocity and coverage clarity is the story.

The LION Lens

What happened — New market research confirms that AI-driven liabilities arising from hallucinations, algorithmic bias, model drift, and data supply chain compromises lack a natural home in any single traditional policy line.

Why it matters — Not all six lines face equal pressure. D&O and EPLI are the most active, with securities class actions and state algorithmic accountability laws (NYC Local Law 144, Colorado, Illinois) creating private rights of action. E&O has filed cases but no definitive verdicts on the professional/product liability boundary. Cyber is stable but contains specific gaps: AI-generated phishing at scale, deepfake social engineering for wire fraud, and model poisoning each test different policy sections, and many policies still define "hacker" as a person. Fiduciary and Crime/FI Bond have real but largely theoretical court exposure.

Practical implications — ISO's new standard-form exclusions (CG 40 47, CG 40 48) are filed and available for carrier adoption. AIG, Great American, and WR Berkley have reportedly taken steps to be able to limit AI exposure on select FI lines. Vendor contracts compound the problem: terms typically cap liability at 12 months of fees with no performance warranties.

So what?

The coverage window is closing on a rolling basis. Clients renewing in 2026 without documented AI governance may face narrower terms or higher retentions. Some carriers are beginning to incorporate governance documentation as a condition of coverage, shifting from treating it solely as an underwriting consideration. We’re still in the early innings, but accelerating.

The LION POV

Here are actions we believe merit consideration based on current market conditions:

• Commission a line-by-line silent AI audit before your next renewal. Map each policy against the six exposure categories. Identify where language is silent, where exclusions have been introduced, and where coverage is affirmed.
• Build your AI governance documentation now. An AI oversight policy, a bias audit schedule, an employee acceptable-use policy, and an incident response playbook aren't compliance exercises. They are coverage positioning.
• Review your AI vendor contracts for liability allocation. If your vendor caps indemnification at 12 months of fees with no performance warranties, your organization is the named defendant.

A note for MGAs: if you deploy AI under binding authority, a biased output creates a liability chain running through your BAA to your capacity provider. Ensure governance documentation satisfies both program underwriters and capacity partners.

Monday morning action: Pull every policy in your current program and flag any line where the word "artificial intelligence" does not appear in either the coverage grant or the exclusions. That's your silent AI exposure map.

(sources: Gallagher Re / MIT / Testudo; DAC Beachcroft; Kennedys Law)

Want to discuss how silent AI affects your institution's program? Contact LION Specialty for a confidential review.

The Market's Answer Is Arriving — A Roundup of Standalone AI Insurance Products

Summary

While some carriers retreat, others are building.

The AI liability insurance market has produced roughly 6–10 distinct standalone products in the last 12–18 months. Armilla AI became the first Lloyd's Coverholder dedicated exclusively to AI liability, covering hallucinations, model drift, agentic AI failures (autonomous systems acting without human approval), and regulatory violations. Testudo, also a Lloyd's Coverholder, launched in January 2026 with a litigation-triggered product for mid-to-large enterprise GenAI deployers. Munich Re offers aiSure performance warranties, HSB launched SMB-focused coverage in March 2026. And Relm introduced three AI-specific products. Google Cloud partnered with Beazley, Chubb, and Munich Re on a tailored cyber solution with affirmative AI coverage for its platform customers. AXA XL, Coalition, and Beazley have added AI endorsements to existing policies.

On the other side of the ledger, AIG, Great American, and WR Berkley filed to restrict AI liability exposure entirely, and the gap between the excluders and the affirmers is exactly where your next renewal conversation lives.

(sources: LION Deep Research; Armilla AI; Testudo; Relm Insurance; Munich Re; Google Cloud; AXA XL; Coalition; Beazley)

So what?

For FI clients, the most relevant products might be Armilla (enterprise focus, two-step underwriting requiring model quality assessment) and Relm's PONTAAI (excess AI wrap for sophisticated deployments). Testudo appeals to institutions prioritizing defense cost coverage without pre-deployment complexity. FI clients already on Google Cloud infrastructure should evaluate the Beazley/Chubb/Munich Re partnership, which offers affirmative AI coverage tied to that platform.

Recent market research flags a risk most product discussions miss…a single failure in a widely used AI system could trigger claims across many unrelated policyholders simultaneously. Reinsurers are developing frameworks to monitor this concentration, but the modeling is early. And no standalone AI product yet has meaningful loss experience to validate its pricing, similar to early cyber market volatility circa 2014–2017.

The standalone AI insurance market is where cyber was in its earliest days. Nascent, fragmented, and about to matter enormously.

Monday morning action: Ask us about standalone AI liability products for your program. Request a market survey before your next renewal.

What's Coming in the Next 12–18 Months

Confirmed: ISO generative AI exclusions are filed and available. Carriers are attaching AI governance questionnaires to D&O, cyber, and EPLI renewals.

Directional: More carriers are expected to introduce AI restrictions on FI lines through 2027. Standalone products will continue entering the market, but capacity depth is unproven. State algorithmic accountability legislation is expanding. The NAIC's model bulletin on AI governance in insurance adds a regulatory compliance layer for carrier clients.

Not yet confirmed: A Lloyd's-style market-wide mandate requiring explicit AI coverage clarity has not been issued. Reinsurance treaty language addressing AI risk is under review but has not produced standardized forms.

Decision timeline: If your next major renewal falls in H2 2026 or H1 2027, the preparation window is now. Audit, governance build, and product evaluation each require 60–90 days of lead time.

The Bottom Line

Silent AI is the single most consequential coverage development facing FI boards in 2026, and most programs are unprepared. Litigation is accelerating, coverage is contracting, and the standalone market is still finding its footing. Institutions that map their exposure and document their governance before renewal will have options. Those that wait could discover their coverage position at claim time.

In Case You Missed It!

Our last Wednesday Intelligence was our most read Brief ever. Worth checking out if you haven't already. It told the story of a southeastern regional insurer we took over from a mega-broker.

The title says it all: "Our Process Wasn't Complicated. That's the Point. And Why We Won."

A D&O claim exposed every gap in the generalist team's process. Our alternative: Our process we run for all LION clients. Senior team, 150-day timeline, direct client-to-carrier communication. The fix wasn't proprietary. It was disciplined. If you missed it, it's worth the four-minute read or listen.

[Listen here / Read here]
​
​Thank you for reading today's edition!

Want to share this edition via text, email or social media? Simply copy-and-paste the link below:

​http://lionspecialty.ck.page/the-single-most-consequential-coverage-issue-facing-financial-institution-boards-in-2026

And if this briefing was forwarded to you, subscribe directly here.
​
​Stay Covered Out There Y'all,

TASH & FLIP

Co-Founders and Managing Partners

LION Specialty
​
P.S. Boards get exposed by D&O mistakes they never see coming. We built a 5-day email course on what to watch for. Comment BLUEPRINT and we'll send it.
​​
P.S.S. Nothing in this briefing constitutes legal advice. These are the opinions of the founders. It's market intelligence designed to help you ask better questions of your advisors and make sharper decisions at your next insurance renewal.