AI Liability Insurance: Frequently Asked Questions

AI liability insurance covers claims that arise when a company's use of artificial intelligence harms someone else, such as a customer misled by a chatbot, a job applicant screened out by an automated tool or an investor who says the company overstated its AI. Affirmative AI coverage names AI in the policy and states how the policy responds, rather than leaving the question to argument after a loss. Mayflower Specialty writes it across directors and officers liability, employment practices liability and professional liability, plus a difference-in-conditions excess layer.

Any company whose AI can affect customers, employees, investors or clients has a reason to look at AI liability insurance, whether it builds its own models or deploys tools bought from a vendor. The exposure is greatest where AI makes or shapes decisions that matter to people, such as hiring, lending, pricing and customer service, and where the board speaks publicly about the company's AI. A company in that position should check how its D&O, EPL and E&O policies treat AI and, where they are silent or exclude it, consider affirmative cover.

Companies already face AI claims before tribunals, courts and regulators. A Canadian tribunal held Air Canada liable for its chatbot's wrong fare advice on February 14th 2024, a federal court preliminarily certified a nationwide age discrimination collective over AI applicant screening in Mobley v. Workday in May 2025, and the SEC settled its first AI-washing cases on March 18th 2024. A public tracker counted more than 2,100 court decisions involving AI-hallucinated content as of October 2026. Because these claims span professional, employment and management liability, each of those policies should be checked for how it treats AI.

Silent AI is the uncertainty that arises when an insurance policy neither covers nor excludes losses caused by artificial intelligence. Many D&O, EPL and E&O wordings were drafted before generative AI came into common use, so whether they respond to an AI claim is left to argument after the loss. Insurers are now settling that silence in one of two directions, by adding AI exclusions or by offering affirmative AI coverage, and a company should know which way its own program is moving before the next renewal.

An AI exclusion is policy wording that removes cover for claims connected to artificial intelligence, and insurers have begun adding them to liability forms. In 2025 W. R. Berkley introduced an “absolute” AI exclusion for D&O, E&O and fiduciary liability that reaches claims “based upon, arising out of, or attributable to” any actual or alleged use of AI, and Verisk's ISO generative AI exclusions for general liability took effect in January 2026. Broad “arising out of” language can catch claims with only a loose link to AI, so renewal wordings deserve a close read.

Cyber insurance responds mainly to security events, such as a breach, ransomware or a network outage, while AI liability insurance is written for the harm an AI system's output or decision causes to someone else, with no breach required. A hiring tool that screens out older applicants, or a chatbot that gives a customer wrong advice, may never meet a cyber policy's trigger. The two policies are designed to sit beside each other, so a company with AI in production should keep its cyber cover and check separately how its liability lines treat AI.

An existing D&O, EPL or E&O policy may respond to some AI claims, but the answer depends on its wording and on any exclusions added at renewal. A policy that is silent on AI leaves room for a coverage dispute, and a policy with a broad AI exclusion can remove cover for any claim that “arises out of” AI. The practical step is to ask your broker for the current wordings, look for definitions and exclusions that mention artificial intelligence, algorithms or automated decisions, and decide whether affirmative cover or a DIC layer is needed.

Mayflower Specialty provides affirmative AI liability coverage in four modules: AI Directors and Officers Liability (AI-D&O), AI Employment Practices Liability (AI-EPL), AI Professional Liability (AI-E&O) and AI DIC Excess. The first three are written for AI claims against the board and officers, against the company as an employer and against its professional services and products, while AI DIC Excess sits over an existing program where those policies are silent on AI or exclude it. Coverage is written on A- (Excellent) AM Best rated paper and is subject to the policy terms.

Each Mayflower module is written for a different kind of AI claim: AI Directors and Officers Liability (AI-D&O) for claims against the board and officers over the company's use, oversight or disclosure of AI, such as AI-washing suits; AI Employment Practices Liability (AI-EPL) for claims over AI used in hiring and other workforce decisions; AI Professional Liability (AI-E&O) for client and third-party losses caused by an AI-enabled product or service; and AI DIC Excess for AI claims where the company's existing policies are silent on AI or exclude it. Whether a particular claim is covered depends on the policy wording and the facts.

AI DIC Excess is a difference-in-conditions layer that sits over a company's existing D&O, EPL and E&O policies and adds affirmative AI coverage where those policies are silent on AI or exclude it. It lets a company keep the program and the insurers it already has while addressing the AI gap that recent exclusions and silent wordings leave. How the layer attaches and responds in a given program is set by the policy terms, so the underlying wordings should be reviewed together with the DIC wording.

Mayflower's AI coverage can be arranged in two ways: as a modular primary policy built from the AI-D&O, AI-EPL and AI-E&O modules a company needs, or as an AI DIC Excess layer over the D&O, EPL and E&O program it already carries. The first suits a company that wants its AI exposure insured on dedicated terms, and the second suits one that is content with its current program apart from AI. A broker can set both options against the existing tower before the company decides.

A claims made and reported policy covers a claim only if it is first made against the insured during the policy period, or any applicable extended reporting period, and is reported within the time the policy requires. Mayflower's AI liability coverage is written on this form. The practical consequence is that timing matters as much as the facts, because a claim reported late can fall outside cover even though the conduct behind it happened while the policy was in force.

Defense costs reduce the limit of liability under Mayflower's policy form, and they are also applied against the retention. Every dollar spent defending a claim lowers the amount left for a settlement or judgment, and heavy defense spending can exhaust the limit altogether. AI claims can turn on long technical disputes about how a model was built, tested and used, so a company should size its limit for the cost of a full defense as well as for the likely settlement.

Mayflower's application places five uses of AI outside its underwriting appetite: autonomous weapons or military targeting, social scoring or mass surveillance, biometric identification in public spaces, real-time emotion inference in workplace or educational settings and deepfake generation without disclosure. Predictive policing and criminal sentencing applications are considered only on bespoke terms. A company whose AI comes close to one of these lines should raise it with its broker, or with Mayflower directly, before applying so the question is settled early.

Mayflower's coverage is built for businesses, so it is not the right fit for individuals or consumers. Nor is it written for first-party security losses such as breach response, ransomware or business interruption, which belong to a cyber policy that Mayflower's AI coverage is designed to sit beside. Companies that use AI for purposes outside Mayflower's underwriting appetite, such as social scoring or undisclosed deepfakes, also fall outside its scope, and it is best to raise any doubt about fit with a broker or with Mayflower before preparing an application.

A company or its broker can start Mayflower's application online, where the questions mirror the AI Liability Supplemental Application and progress is saved along the way, or it can download the PDF version after accepting its license terms. The supplemental application is completed alongside the standard ACORD applications for D&O, EPL and professional liability. Once a submission arrives, Mayflower's underwriting team responds within 48 hours.

Mayflower places its coverage through brokers, so a company buying a policy will do so through its broker. A company can still start the online application itself, which lets the underwriting conversation begin while the broker reviews the rest of the program, and it can also contact Mayflower with coverage questions before choosing to apply. Bringing the broker in early helps, because the broker holds the current wordings and renewal dates that the AI coverage has to fit around.

Mayflower's AI Liability Supplemental Application has 11 sections: applicant information; AI systems overview; AI governance; data governance; system operations and monitoring; AI incident response; regulatory environment and compliance; claims and loss history; prior and current insurance; required documentation; and applicant representation and warranty. The sections on AI systems, governance, data, operations, incident response and regulation describe what the company's AI does and how it is controlled, which is the basis of Mayflower's underwriting, so it helps to prepare them with engineering, legal, risk and HR together.

Mayflower's application requires five documents: completed ACORD application(s) for the applicable coverage lines; three years of loss runs for D&O, EPL and E&O; most recent annual financial statements or audited financials; AI governance policy or responsible AI framework; and AI system inventory (systems, use cases and data types). Other documents are recommended rather than required and, in the application's words, “may improve terms”, including board or committee minutes on AI risk oversight, model cards, bias audit or fairness testing results, an AI incident response plan and an ISO 42001 certificate or audit report where one exists.

Mayflower underwrites on the applicant's AI governance, meaning how its AI decisions are made, monitored, documented and escalated. Documented controls such as an AI system inventory, human review of consequential decisions, bias testing and an incident response plan make a claim less likely or less severe, and the application states that recommended governance documents may improve terms. Price and terms still depend on the exposure and the claims history, so no particular credit for governance can be promised in advance.

Submitting Mayflower's application does not bind coverage, and the application says so in its representation and warranty section. The application gives the underwriter what it needs to prepare a quote, and the company decides whether to proceed once terms are offered. If a policy is later issued, the application becomes part of it, which is why the signing officer declares the answers true and complete and the company must report any material changes that arise before the policy is issued.

Mayflower's application states that all information and submitted materials are held in confidence. Applicants often share sensitive material, such as an AI system inventory, bias audit results or board minutes, and that material is used to assess the risk. The application also notes that Mayflower may supplement it with publicly available information for risk assessment, so the company's public statements about its AI, such as filings and marketing claims, should be consistent with its answers.

Mayflower places its coverage through brokers. A broker can submit a risk by starting the online application in the broker role or by sending the PDF supplemental application completed with the client, together with the ACORD applications and the required documents. The answers still need input from the client's own teams, and the client's Chairman, CEO, CFO, President or General Counsel must sign. Brokers who are not yet appointed with Mayflower can request an appointment through the contact form.

Claims under Mayflower's AI liability coverage are handled by a specialist claims administrator. AI claims raise technical questions that ordinary liability claims often do not, such as what a model did, what data it relied on and who controlled it, so it helps to have them handled by people who understand how AI losses develop. The policy sets out where and how notice of a claim must be given, and it is worth confirming those steps with your broker when the policy is bound rather than after an incident.

An AI incident that might lead to a claim should be reported as soon as practicable and within the time the policy requires, because a claims made and reported policy covers only claims made and reported within the periods it sets. Report through your broker, follow the notice provision in the policy and keep a record of what happened, which systems were involved and when the company first learned of it. Mayflower's application also warns that failing to tell a current insurer about known claims or circumstances before that policy expires may create a lack of coverage.

Whether a specific AI claim is covered depends on the policy wording, the facts of the claim and the information given in the application, so no broker or underwriter can promise an outcome in advance. Affirmative coverage changes the starting point, because the policy names AI exposures and states how it responds instead of leaving the question to an argument over silent wording. Reading the insuring agreements, definitions and exclusions with your broker before a loss is the best way to know where you stand.

A claim first made after a claims made and reported policy ends is generally outside that policy unless an extended reporting period applies or a renewal policy picks it up. An extended reporting period gives extra time to report claims arising from conduct before the policy ended, and its availability and terms are set by the policy. Keeping cover continuous, and checking that the retroactive date on any new policy reaches back far enough, avoids gaps when a company changes its program.

Mayflower Specialty is an AI liability insurance specialist that provides coverage for companies that build and deploy AI. It writes affirmative AI coverage for directors and officers liability (AI-D&O), employment practices liability (AI-EPL) and professional liability (AI-E&O), plus an AI DIC Excess layer over an existing program, on A- (Excellent) AM Best rated paper backed by some of the world's largest reinsurers. Coverage is placed through brokers and underwritten on the applicant's AI governance.

Mayflower's coverage is written on A- (Excellent) AM Best rated paper and is backed by some of the world's largest reinsurers, whose names are confidential, and claims are handled by a specialist claims administrator. For a buyer, the rating of the paper is the main published measure of the financial strength behind a policy, so it is worth confirming the issuing company and its current rating with your broker when terms are offered.

An AM Best Financial Strength Rating is AM Best's independent opinion of an insurer's ability to meet its ongoing insurance obligations. Ratings of A and A- form the “Excellent” category, which AM Best assigns to insurers that have, in its opinion, “an excellent ability to meet their ongoing insurance obligations”, and which sits second on its scale after “Superior” (A++ and A+). AM Best notes that the rating addresses financial strength and is not a recommendation of any particular policy.

Jeremy Epstein founded Mayflower and serves as its CEO. The company has raised funding from Clocktower Ventures, Antler and angel investors from the insurance and financial services industry, and it was selected for the FinTech Innovation Lab New York 2026 class, a 12-week program established by Accenture and the Partnership Fund for New York City.