By the Editors · Published 14 June 2026 · Last reviewed 14 June 2026 · 12 minute read

Who insures AI agents in Europe?

Every month the AI insurance market shifts. New programmes are announced, funding rounds close, and brokers update their market guides. This reference page maps who is actually writing or developing AI agent coverage in Europe as of mid-2026, what each programme genuinely is (licensed carrier, MGA, guarantee, or roadmap), and what an SME operator can and cannot buy today.

The state of the market in plain terms

If you search "who insures AI agents in Europe" in mid-2026, you will find three types of results. The first is companies selling their own products. The second is enterprise-oriented programmes that require a formal underwriting submission and are not accessible to most SMEs without a specialist broker. The third is roadmap announcements from companies building toward the market rather than operating in it.

What you will not find is a neutral map of who is what. That is what this page provides. Future Proof Intelligence is not a licensed insurance broker or carrier. We publish this as an independent reading and readiness layer above the market. No entry below is sponsored or paid. Every entry below is built only from what the provider itself publishes. Where a provider does not publish its underwriter, its limits, its pricing or its geographic scope, this page says so rather than estimating it.

Programme by programme: who is actually writing or developing AI agent cover

Munich Re aiSure

What it is: A performance guarantee for AI systems, settling parametrically on proven AI errors rather than through individual loss adjustment. A reference programme in the space.

What it covers: AI model performance failures. Covers losses an AI makes that a human would not, measured against a defined performance baseline.

European availability: Yes, through Munich Re's European operations. Enterprise-focused; no SME pricing tiers announced.

Why it matters: aiSure is the longest-established named AI risk product from a tier-one reinsurer. Understanding how it settles clarifies the performance-guarantee model that parts of the broader market are building on.

SME accessibility: Low. Built for enterprise AI deployments with measurable performance data. Most SME operators do not have the data infrastructure or governance documentation aiSure requires.

AIUC-1 standard and the ElevenLabs policy

What it is: AIUC-1 is a certification and assurance standard for AI agents published by the Artificial Intelligence Underwriting Company (AIUC). It is not an insurance product itself; it is the standard against which an agent is assessed before insurance is written.

What happened: On 12 February 2026, ElevenLabs became the first company to have an AI agent deployment insured against AIUC-1. The policy covered voice agents and was underwritten through Lloyd's of London. AIUC-1 certification subjects a system to more than 5,000 adversarial simulations across security, safety, reliability, data and privacy, accountability and societal risk.

European availability: AIUC-1 is a global standard and AIUC states that certified coverage is placed through Lloyd's of London. AIUC publishes no list of distribution partners, so a European operator cannot yet route to a named local intermediary.

Why it matters: The ElevenLabs policy established the precedent that structured certification evidence is the key to binding AI agent coverage. Every programme in this list is, in some sense, a variation on or competitor to this model.

Armilla

What it is: An AI assurance firm and Lloyd's of London coverholder. Capacity providers: Chaucer Group, Axis Capital and Convex.

What it covers: Structured AI liability underwriting for technology companies and enterprises. Armilla does not publish standard limits; they are agreed per submission.

European availability: Yes, through Lloyd's. Available to European technology companies with a formal underwriting submission.

Pricing: Not published. Armilla underwrites from a submission and requires governance documentation.

SME accessibility: Moderate. Armilla targets technology companies rather than all SMEs. Requires a detailed submission. Not a click-and-buy product.

Klaimee

What it is: A YC-backed startup describing its product as purpose-built E&O for AI agents. Covers eight risk dimensions aligned to EU AI Act framing.

What it covers (as described by Klaimee): Liability arising from agentic AI that traditional errors and omissions and cyber policies carve out. Klaimee's published process is a public data scan, a governance assessment and behavioural testing, with a certification score that drives the premium.

European availability: Klaimee publishes no geographic scope and names no underwriter. Operators in Europe should request written confirmation of who carries the risk, and under which regulator, before purchasing.

SME accessibility: Not stated. Klaimee publishes no pricing and no limits.

Mount

What it is: A Y Combinator backed startup (Spring 2026) that describes itself as an AI agent insurance carrier. That description is Mount's own; this publication has seen no filing or regulator entry confirming carrier licensing, and Mount publishes none.

What it covers: Mount's published product, ActiveCover, pairs continuous vulnerability scanning and red teaming with liability cover for AI agent workflow failures.

European availability: Mount publishes no European distribution or licensing detail. Treat it as US first until it does.

SME accessibility: Mount describes its first product as built for mid-market and enterprise companies, not SMEs.

Agensure

What it is: A European company that runs adversarial testing against AI agents, publishes an Agensure Risk Score (ARS, 1 to 100), and issues an Agent Deployment Readiness (ADR) certificate with a QR code linking to live verification. It maps its certificate profile to EU AI Act Article 15 on accuracy, robustness and cybersecurity.

What it covers: Nothing, in the insurance sense. Agensure is not an insurer and publishes no insurance product. What it sells is adversarial testing and a certificate.

European availability: Yes, as a testing and certification service. There is no policy to buy.

Why it matters: Agensure is the most European-native programme in this list and frames its certificate against EU AI Act Article 15. Adversarial test evidence of this kind is exactly what the certification-led underwriting model asks for, so its output may become an underwriting input as the European market develops.

HSB AI Liability Insurance

What it is: A product launched in the US in March 2026 by HSB (Hartford Steam Boiler), a Munich Re subsidiary. An SME-facing AI liability product from a major carrier.

What it covers: Three categories of third-party liability from AI use: bodily injury from AI-controlled systems, property damage from AI-generated instructions, and personal and advertising injury from AI-generated content.

Pricing: Not disclosed at launch. Limits and deductibles sit inside the partner carrier's programme.

European availability: None announced. HSB's launch was United States only, distributed through partner carriers subject to state-level regulatory approval.

What the comparison table shows

Why every result you find is a vendor selling itself

The honest answer to "who insures AI agents in Europe" is that the market is being built in public. Every programme in this list has an incentive to position itself as the default answer. None of them would describe themselves as early-stage, enterprise-first, or US-distributed if they could avoid it.

This creates a specific problem for SME operators trying to make a genuine procurement decision. The information available through search results and vendor pages describes a market that is more mature than it actually is. The programmes that exist are real. The coverage they provide is real. But the framing around European SME accessibility is consistently ahead of what you can actually buy.

The most useful thing an SME operator can do is not to pick a provider from a list but to build the evidence file that any of these programmes will require. Every programme above needs the same four inputs: documented scope, human oversight procedures, AI risk assessment, and incident response. An operator who has those four documents is ready to have a real conversation with any of the brokers who can access these programmes. An operator who does not have them will not receive a binding indication regardless of which programme they approach.

The certification layer that connects to coverage

The Agent Certified methodology at agentcertified.eu was built specifically to produce the evidence file underwriters increasingly require. Its seven dimensions (autonomy envelope, data governance, human oversight, security and resilience, performance and reliability, transparency and trust, and governance) map to the underwriting criteria used by the programmes in this list.

This is not marketing. The AIUC-1 standard and the Agent Certified methodology address the same underlying question: can this AI agent deployment be described, documented, and assessed against a published framework? If the answer is yes, coverage becomes negotiable. If the answer is no, it is not.

What to do this month

If you are an SME operator trying to find AI agent insurance in Europe in mid-2026, three steps move you forward regardless of which programme ultimately suits your situation.

First, run the coverage audit at tools/coverage-audit. It takes fifteen minutes and tells you where your existing policy gaps are. That framing is what you bring to a broker conversation.

Second, send a written request to your broker asking for a position on each of your current policies relative to AI agent risk. Do not accept a verbal answer. Ask for clause references from the most recent renewal.

Third, begin the Agent Certified assessment at agentcertified.eu. The seven-dimension framework produces the evidence file that all the programmes above require before they will quote you. Starting this now means you are ready to move when a product becomes accessible to you.

The get covered page walks through the full pathway from diagnostic to carrier conversation in detail.

Frequently asked questions

Who insures AI agents in Europe in 2026?

As of mid-2026, no licensed European-native carrier sells off-the-shelf AI agent liability cover to SMEs at scale. The options are Munich Re aiSure (AI performance cover, enterprise), Armilla (Lloyd's coverholder), AIUC-1-anchored coverage (established by the ElevenLabs policy in February 2026), Klaimee and Mount (United States based, no published European distribution), Agensure (EU testing and certification service, not an insurer), and HSB AI Liability (United States only, no European availability announced).

Is there a licensed European insurer for AI agents?

Not one selling off-the-shelf SME cover as of mid-2026. Armilla operates through Lloyd's of London, which is a regulated UK market. Agensure is EU-based but is an assessment service, not a carrier. The market is developing rapidly and this assessment will need updating quarterly.

What is the cheapest AI agent insurance available in Europe?

HSB's product is SME-facing AI liability cover from a tier-one reinsurer subsidiary, but it disclosed no pricing or limits at launch and announced no European availability. For a European operator today, the most immediately available route is not a dedicated product at all: it is an AI endorsement negotiated onto an existing professional lines policy at renewal. That is not the same cover, and the additional premium is set case by case rather than from a published rate.

What will the European AI insurance market look like by 2027?

The most likely scenario is that two or three of the programmes in this guide will develop accessible SME products by the end of 2027, driven by the volume of operators who have completed some form of AI governance assessment and the pressure of the revised Product Liability Directive (applying from December 2026). The programmes with reinsurance infrastructure in place (Munich Re, Lloyd's) are the most likely to move quickly once demand and evidence quality justify it.