For Brokers

The Broker Hub for AI insurance.

Your clients are deploying generative AI. Their E&O and CGL forms exclude third-party harms caused by that AI. Testudo writes the standalone third-party liability policy that fills the gap.
Underwritten by certain underwriters at Lloyd's.
Lloyd's Coverholder
A-rated capacity
ApolloAtriumQBEHiscox
  1. 01

    What is generative AI?

    Generative AI refers to models that produce new content (text, images, video, audio, or code) in response to a prompt. ChatGPT by OpenAI and Claude by Anthropic are the best-known examples, and they're the same systems your clients are already running.

  2. 02

    Your insureds are already using AI today.

    50%+

    of U.S. organizations have deployed AI in at least one business function

    From content generation to website chatbots, organizations across most industries are folding generative AI into how they operate. The accounts you'll renew next month are already using AI in production.

  3. 03

    Liability is rising, and coverage is being taken away.

    140%+

    year-over-year growth in AI-related U.S. lawsuits filed

    ISO AI exclusions have been filed into General Liability forms, and similar carve-outs are spreading into E&O and Professional Lines. As the claims land, insurers will be excluding this exposure.

  4. 04

    Get appointed. Distribute AI insurance.

    Get appointed

    Standard producer agreement, no volume minimums, onboarded in 3-5 days. Your first submission can be the same week.

    Book a webinar

    Live session where we walk your team through AI exposure, coverage, and how to sell AI insurance. Virtual or in-person.

Third-Party Gen AI Liability Cover

What does the policy cover?

Five insuring agreements purpose-built for generative AI deployer exposure.

Our policy responds when the outputs of the generative AI system the insured has deployed cause harm to third parties. Coverage triggers and definitions are written for how generative systems actually fail in production, not bolted onto a tech E&O form.

Generative AI Errors

Financial loss suffered by a third party, arising from negligent errors or misstatements caused by a generative AI output. For example, the hotel chain's chatbot that quotes a rate the brand can't honor.

IP Infringement and Personal Injury

Infringement of copyright, trademark, libel, and slander arising from a generative AI output. For example, the advertising agency whose AI-generated campaign reproduces a competitor's trademarked design.

Unauthorized Data Disclosure

Liability arising from the unauthorized disclosure of Protected Information through a generative AI output. For example, the hospital's chatbot that reveals one patient's medical records to another.

Bodily Injury

Bodily injury to a third party who relied on a Gen AI output. For example, the worker an AI site coordinator routed into a live lift zone.

Property Damage

Physical damage to third-party property caused by action taken on a Gen AI output. For example, the slab cracked under a lift the AI cleared.

Additional Coverage Areas

Coming soon

Limits & pricing
$2MDefault per insuring agreement
$10MAggregate across all five
$1M-$10M+Limit range
  • Our claims-made coverage is distributed on an excess-and-surplus-lines basis.
  • Lloyd's provides A+ (Superior) A.M. Best-rated capacity.
  • Higher limits available case-by-case for qualified risks.
  • Pricing is provided on our bespoke models.
  • Governing law is New York.

WHO'S IT FOR?

Our target customers.

Any organization up to $10B in revenue that has deployed generative AI in its operations.

Your deployer clients sit at the end of the AI value chain. They don't build the models, they operate them. That's where the exposure sits.

Developers

The developer creates the foundational model, such as GPT-5 or other large language models. They own the core training, architecture, and base functionality. Examples include OpenAI, Anthropic, Google, xAI, Cohere, and Mistral.

Vendors

The vendor customizes or modifies the foundational AI model to create a specific generative AI system or application (e.g., a chatbot) and sells it to deployers. Examples include Cursor, Harvey AI, Bland.AI, and Microsoft Co-Pilot.

Deployers

The deployer embeds a developer's or vendor's AI system into its operations as customer chatbots, internal copilots, and workflow automation. Our target customers: any commercial company running Gen AI in production, from manufacturers and retailers to healthcare and professional services.

Conversation starters

Surface the exposure on the next renewal call.

Most deployers won’t raise their AI use on their own. It reads to them as an IT question, not an insurance one. Three questions surface it on the renewal agenda. A yes on any two is enough to submit.

  1. 01

    Are you currently using any generative AI, and if so, for what specific functions?

    Generative AI systems include chatbots, support copilots, and virtual assistants.

  2. 02

    How much visibility do your risk team and legal have into the specific Gen AI tools your business units are running today?

    Ungoverned AI is where exposure compounds. When business lines deploy without risk-team visibility, the liability still lands on the insured.

  3. 03

    Have you confirmed with your current carriers that AI-related losses are still covered on your E&O, Cyber, and CGL placements this renewal?

    Generative AI risk sits in a gap between incumbent forms. E&O, Cyber, and CGL weren't scoped for it, and carriers are now excluding it at renewal.

APPETITE

What's in appetite, and what isn't.

Where we lean in, and what we ordinarily decline, by Gen AI use case and industry.

The lists below aren't exhaustive. They set expectations. If a client's use case or industry isn't listed, bring it to us and we'll handle it case-by-case.

In appetite

Use cases

We have the appetite to review Gen AI deployed across these use cases:

  • Chatbots (customer support and success, companion chatbots)
  • Inbound/outbound calling
  • Embedded AI in product or safety features
  • Virtual meeting transcription
  • Pharmaceutical discovery
  • Creative output in advertising, design, and entertainment
  • Advertising and marketing content generation
  • Sales, CRM, and client acquisition
  • Procurement & vendor management
  • Personal and team efficiency (virtual executive assistants)
  • Data science and analytics
  • Internal compliance
  • Cyber security and threat detection (inc. fraud)
  • Supply chain

Industries

We have the appetite to review Gen AI deployed in these industries:

  • Accommodation and Food Services
  • Administrative and Support Services
  • Arts, Entertainment, and Recreation
  • Construction
  • Educational Services
  • Finance and Insurance
  • Health Care and Social Assistance
  • Information
  • Manufacturing
  • Mining, Quarrying, Oil and Gas
  • Professional, Scientific, Technical Services
  • Real Estate, Rental and Leasing
  • Retail Trade
  • Transportation and Warehousing
  • Utilities
  • Wholesale Trade
Out of appetite

Use cases

We ordinarily decline Gen AI deployed for these use cases:

  • Facial recognition
  • Emotional inference in the workplace
  • Social scoring
  • Biometric categorization & identification
  • Law enforcement & crime prediction
  • Migration & border control
  • Critical infrastructure
  • Administration of democracy & justice
  • Military applications
  • Adult content
  • Political advertising
  • Counselling & mental health
  • Automated lending & payment processing
  • Gambling
  • Hiring
  • Gaming
  • Education (learning or marking)

Claims Examples

A scenario for your book.

Reach out and we will share claims scenarios across industry verticals most relevant to you, based on real-world data, litigation, and incidents.

ScenarioTAI-S-2025-047
Hypothetical
Industry
Auto Finance
Failure
APR hallucination
Filed
08 Feb 2025
Coverage
01Generative AI Errors
████████████████████████████████
████████████████████████
████████████████████████████
Request full scenario
ScenarioTAI-S-2025-073
Hypothetical
Industry
Habitational
Failure
Bodily injury
Filed
18 Aug 2025
Coverage
04Bodily Injury
████████████████████████████████
████████████████████████
████████████████████████████
Request full scenario
ScenarioTAI-S-2025-065
Hypothetical
Industry
Manufacturing
Failure
Trademark infringement
Filed
03 Jul 2025
Coverage
02IP Infringement and Personal Injury
████████████████████████████████
████████████████████████
████████████████████████████
Request full scenario
ScenarioTAI-S-2025-091
Hypothetical
Industry
Construction
Failure
Property damage
Filed
22 Sep 2025
Coverage
05Property Damage
████████████████████████████████
████████████████████████
████████████████████████████
Request full scenario
ScenarioTAI-S-2025-118
Hypothetical
Industry
Hospitality
Failure
Guest data leak
Filed
14 Nov 2025
Coverage
03Unauthorized Data Disclosure
████████████████████████████████
████████████████████████
████████████████████████████
Request full scenario

HOW IT WORKS

How does the policy work?

Generative AI Deployed

The insured uses a generative AI tool such as a website chatbot.

Harmful Output

The generative AI tool produces an output that causes harm to a third party.

Claim Filed

The third party who was harmed sends a written demand or files a suit.

Insurance Responds

Testudo's policy responds to cover defense costs and settlements up to the policy limit for covered claims.

Submitting a risk
2.5Pages

A concise, broker-friendly intake covering the deployer's profile and the Gen AI systems in scope.

< 3 minTime to complete

High-level questions designed to be answered alongside the client in a single discovery call.

0Audits, tests or evals

No security audits, penetration tests, or AI evals. Our application form is typically enough to price the risk.

Portfolio Review

We'll walk through your book with you, one account at a time, to map where AI exposure lives and where there's opportunity for cover.

You share a list of the named insureds in your book. We handle the classification, exposure grading, and appetite review. Most books come back within two business days of the list landing with our underwriting team.

Portfolio_Analysis · Sample broker book
A
B
C
D
E
1
Account
Gen AI use case
Exposure
Engage?
UW notes
2
Harbor Supply Co.Industrial Distribution
Customer support chatbot
Material
Yes
GAI Errors
3
Meridian Retail GroupRetail Trade
Marketing copy generation
Moderate
Yes
IP / PI
4
Bayline Health SystemHealthcare
Patient scheduling agent
Material
Yes
Unauth. Data
5
Crestwood LogisticsTransportation
Dispatch copilot
Low
Watch
Under review
6
Quadrant Consumer FinanceFinance & Insurance
Automated loan underwriting
Material
No
Out of appetite
5 of 23 accounts reviewed. Complete analysis on request.

BROKER EDUCATION

Book a webinar with our team.

We'll fly out or jump on video to educate your producers on AI exposure, and how to sell AI insurance with confidence.

  • VirtualZoom, Teams, Meet
  • In-personWe'll come to you

Gap Analysis

Existing policies leave material gaps.

Commercial General Liability, E&O, and Cyber can provide indirect coverage for some Gen AI risks, but none offers comprehensive protection. A purpose-built Generative AI Third-Party Liability policy is designed to fill the gap.

In response to damages from Gen AI outputs

Does the policy respond to third-party financial damages?

  • CGLNo

    CGL does not cover pure economic loss.

  • E&OYes

    Unless Gen AI exclusions are used.

  • CyberNo

    Even if arising from a covered cyber event.

  • Testudo Gen AI TPLYes

    Under the Generative AI Errors insuring agreement.

In response to damages from Gen AI outputs

Does the policy respond to third-party bodily injury and property damage?

  • CGLYes

    Unless Gen AI exclusions are used.

  • E&ONo

    E&O typically excludes BI / PD.

  • CyberNo

    Cyber typically excludes BI / PD.

  • Testudo Gen AI TPLYes

    Under the Bodily Injury and Property Damage insuring agreements.

In response to damages from Gen AI outputs

Does the policy respond to third-party damages arising from unauthorized data disclosure?

  • CGLNo

    CGL often has an access or disclosure exclusion.

  • E&ONo

    E&O typically would not respond.

  • CyberPartially

    Potentially, depending on the wording.

  • Testudo Gen AI TPLYes

    Under the Unauthorized Data Disclosure insuring agreement.

In response to damages from Gen AI outputs

Does the policy respond to personal injury?

  • CGLPartially

    To certain personal and advertising injury claims.

  • E&OPartially

    To a narrow definition of personal injury.

  • CyberPartially

    If personal injury is covered under 'media liability'.

  • Testudo Gen AI TPLYes

    Under the IP Infringement and Personal Injury insuring agreement.

In response to damages from Gen AI outputs

Does the policy respond to inadvertent IP infringement?

  • CGLPartially

    To a narrow advertising-related IP infringement.

  • E&ONo

    E&O excludes IP infringement.

  • CyberPartially

    IP infringement is narrowly defined under 'media liability'.

  • Testudo Gen AI TPLYes

    Under the IP Infringement and Personal Injury insuring agreement.

COMMON QUESTIONS

Frequently asked questions.

  • Scheduling specific systems gives the client certainty of coverage. Blanket policies often lead to coverage disputes.

    The schedule also protects the client from shadow AI, where employees use unsanctioned AI tools without IT approval. We write coverage for the AI systems the client has reviewed and approved. Anything outside that list we don't price, because we don't know it exists. A schedule keeps coverage and premium aligned to what's actually deployed.

    We are happy to add new Gen AI systems to the schedule on request, subject to an underwriting review.

  • Testudo writes US risks today because the US is where the claims environment is most developed. That includes ISO CGL exclusions from January 2026, the Moffatt v. Air Canada precedent, and the bulk of early-stage litigation in federal and state courts.

    Our underwriting data (claims development, deployment patterns, model performance) is mainly US-sourced. We expect to expand internationally as non-US markets reach the same claims maturity.

    For a global client operating worldwide with a US HQ, or any other bespoke risk, please reach out to the team on a case-by-case basis.

  • Commercial companies up to $10B in revenue, from mid-market through large enterprise. You can pitch this across most of your book, including manufacturers, retailers, professional services, banks, and healthcare organizations. If a client is using generative AI internally, in a customer-facing, or decision-making workflow, they are in the target market.

    We supply risk reports to the broker to help start the conversation. They come industry-specific, or company-specific for a named policyholder.

  • Cyber insurance may include a narrow section covering media liability. Testudo's IP Infringement and Personal Injury agreement responds to copyright, trademark, libel, and slander from a Gen AI output in one trigger. Cyber wordings vary widely and we regard that coverage as insufficient for an AI-deployer policyholder. Cyber insurance was not built for Gen AI deployment.

    Where the client already has the coverage under another policy, we can write Testudo on a modular basis, dropping the agreements that duplicate what's already placed and pricing to what's left.

  • No. We exclude cyber-attacks.

    If the loss is caused by a cyber-attack, the cyber insurer pays, subject to the cyber contract. Cyber insurance policies generally respond to the defined 'Computer System.'

  • No. Testudo's Gen AI Third-Party Liability is a commercial-only product.

WAYS TO ENGAGE WITH TESTUDO

Pick your path.