Claim Scenario
Your AI gives a wrong answer that causes a third party financial loss.
- General LiabilityNo
- Tech E&OLimited
- Cyber InsuranceNo
- Testudo AI LiabilityCoveredGenerative AI Errors
Generative AI Liability Insurance
You're responsible for your AI. Your existing coverage may not be.
Many businesses still assume they're covered. But as generative AI lawsuits climb, insurers are responding by excluding generative AI risks from traditional coverage and denying related claims.
Underwritten at Lloyd's
The Solution
Coverage for the ways generative AI creates liability.
As generative AI makes decisions, creates content, and powers critical work, it can also create new paths to liability. To protect your business, we provide liability coverage for the cost of lawsuits caused by your use of generative AI.
01
Hallucination liability
When hallucinations or incorrect outputs cause financial loss to a third party.
02
Intellectual Property
When generative AI outputs infringe on intellectual property rights.
03
Physical injury
When your use of generative AI physically injures a person or property.
04
Privacy
When generative AI outputs reveal protected data about a person without their permission.
AI Hallucination RiskIP Infringement & Personal InjuryUnauthorized Data DisclosureBodily InjuryProperty Damage
AI Hallucination RiskIP Infringement & Personal InjuryUnauthorized Data DisclosureBodily InjuryProperty Damage
What is generative AI liability insurance?
Generative AI liability insurance is a standalone, claims-made third-party liability policy that covers the defense costs and damages a company faces when the generative AI it deploys causes harm to a third party. Testudo writes it for U.S. companies deploying generative AI, underwritten by syndicates at Lloyd's (rated A+ (Superior) by AM Best), on an excess-and-surplus-lines basis in all 50 states.
The policy is purpose-built for the gap the standard market leaves, with five insuring agreements.
Source on the new exclusions: Verisk/ISO general-liability exclusions for generative AI.
- 01
AI Hallucination Risk
Financial loss to a third party from a hallucination or other negligent error or misstatement in a generative AI output (the policy's Generative AI Errors insuring agreement). Example: a customer-support assistant states a price or policy term the business can't honor, and the customer sues for the difference.
- 02
IP Infringement & Personal Injury
Infringement of copyright, trademark, libel, and slander arising from a generative AI output. Example: a marketing tool generates a campaign that uses the likeness of a living model as a false endorsement, and the model sues.
- 03
Unauthorized Data Disclosure
Liability from the unauthorized disclosure of protected information through a generative AI output. Example: a transcription bot inadvertently discloses one customer's account details to another, and the affected customer sues.
- 04
Bodily Injury
Bodily injury to a third party who relied on a generative AI output. Example: a school's AI sports assistant advises an athlete to undertake overexerting physical activity, and the athlete is injured.
- 05
Property Damage
Physical damage to third-party property caused by action taken on a generative AI output. Example: a construction site's AI-generated installation procedure causes surrounding property to be damaged.
Read more about how Testudo's coverage responds to generative AI risks, including AI hallucinations.
What a generative AI claim looks like.
Hypothetical claim scenarios across Testudo's five insuring agreements.
- Enterprise support
Support chatbots for enterprise companies in telecoms, SaaS, and fintech provide clients incorrect onboarding or integration steps, leading to client downtime and lost revenue.
Generative AI errors - Retail
A chatbot fabricates a sexual misconduct case against the CEO of a supermarket.
IP & personal injury - Healthcare
Wellness AI gives unsafe recovery guidance, allegedly causing physical injury.
Bodily injury - SaaS
A helpdesk AI provides a customer with incorrect steps for subscription cancellation and the customer gets charged for another year.
Generative AI errors - Marketing
AI writes a marketing campaign using a tagline that turns out to be a competitor's registered trademark, triggering a legal complaint.
IP & personal injury - E-commerce
A retail chatbot reveals another customer's order history, bank details, and address.
Unauthorized data disclosure - Legal services
Contract-summary AI misstates service terms, causing a missed contractual obligation.
Generative AI errors - Financial services
An AI vendor-risk summary falsely flags a supplier as being on a sanctions list.
IP & personal injury - Real estate
An AI assistant tells a building manager to set the HVAC to the wrong temperature, freezing and bursting the pipes in a tenant's unit.
Property damage
Scenarios are hypothetical and shown for illustration. Coverage determinations depend on the specific policy wording and claim facts.
Wondering if you're covered for claims like these?
Coverage at a Glance
Where your existing policies leave you exposed.
The generative AI exposures your business now faces, mapped against the policies most businesses already carry.
Claim Scenario
Your AI's output infringes a copyright or trademark, or harms a reputation.
- General LiabilityLimited
- Tech E&ONo
- Cyber InsuranceLimited
- Testudo AI LiabilityCoveredIP Infringement and Personal Injury
Claim Scenario
Your AI exposes private information about someone without their permission.
- General LiabilityNo
- Tech E&ONo
- Cyber InsuranceLimited
- Testudo AI LiabilityCoveredUnauthorized Data Disclosure
Claim Scenario
Your use of AI physically injures a person.
- General LiabilityLimited
- Tech E&ONo
- Cyber InsuranceNo
- Testudo AI LiabilityCoveredBodily Injury
Claim Scenario
Your use of AI damages physical property.
- General LiabilityLimited
- Tech E&ONo
- Cyber InsuranceNo
- Testudo AI LiabilityCoveredProperty Damage
Limitednarrow scope or subject to generative AI exclusions.
Speed to Coverage
Covered in days, not weeks.
Getting covered is fast and runs through the broker you already work with. Our AI Risk Engine does the first read and our underwriters make every call, so speed never means lighter underwriting.
- 01
Submit through your broker
Testudo works only through brokers, so cover fits the relationships you already have.
- 02
No audit of your stack
No audits, penetration tests, or AI evals to pass, and nothing for your security team to clear. We work from public signals and what you share, never from inside your systems.
- 03
A fast decision, fully underwritten
A person underwrites every policy, not just a model. You get a fast decision priced to your real exposure, never a lighter one.
- 04
Bound within a week
A price indication within 48 hours, then cover bound within a week of a completed application, on Lloyd's paper rated A+ (Superior) by AM Best.
COMMON QUESTIONS
Frequently asked questions.
It is a standalone, claims-made third-party liability policy with five insuring agreements: Generative AI Errors, IP Infringement and Personal Injury, Unauthorized Data Disclosure, Bodily Injury, and Property Damage.
It responds to the defense costs and third-party liabilities arising from the outputs of the generative AI systems scheduled on the policy.
U.S. companies deploying generative AI in their products or operations, from mid-market through large enterprise. Standard Commercial General Liability, Cyber, and technology E&O policies increasingly carve out generative AI exposure, leaving deployers without affirmative coverage for AI-driven harm (Verisk/ISO filed general-liability exclusions CG 40 47 and CG 40 48 (plus the products and completed operations form CG 35 08), attaching at renewals from January 1, 2026; sourced in the definition above).
No. Cyber and technology E&O insurance were not built for the gen AI deployer, and the narrow media-liability section some cyber wordings include is insufficient when an AI output is the cause of loss. This policy responds to third-party liability from the outputs of your generative AI systems, which cyber and E&O forms increasingly exclude. Where cover overlaps, Testudo can be written modularly, dropping the agreements that duplicate what is already placed and pricing to what is left.
When generative AI causes harm to a third party (an incorrect output, IP infringement, unauthorized data disclosure, or physical injury), liability typically attaches to the enterprise that deployed the system, not the AI vendor that built the model. Vendor contracts and technology E&O policies are designed for the vendor side of that split.
Generative AI liability insurance is written for the deployer side. It responds to the defense costs and damages arising from the outputs of the generative AI systems scheduled on the policy.
Coverage is distributed on an excess-and-surplus-lines basis in all 50 U.S. states, written on Lloyd's of London paper.
Commercial general liability (CGL) policies increasingly do not cover generative AI liability. Insurers are adding ISO/Verisk generative-AI exclusion endorsements (forms CG 40 47, CG 40 48, and CG 35 08) that carve out bodily injury, property damage, and personal and advertising injury arising from generative AI output. Standalone generative AI liability insurance is designed to fill that gap.
Cyber insurance generally covers data breaches, network security failures, and privacy events, not third-party harm from what a generative AI system produces. Claims like defamation, IP infringement, or faulty AI-generated advice typically fall outside cyber policies, and outside CGL where AI exclusions apply, leaving a gap that standalone generative AI liability insurance addresses.
Cover the generative AI risk on your balance sheet.
Tell us what your business is deploying and we will show you how a standalone Gen AI third-party liability policy responds where your current tower falls short.