Aipedia: Actuaries
Actuarial PI usually responds to negligence claims whether or not AI was involved, subject to the policy terms, because it covers the professional service rather than the tool. The detail matters: exclusions, how work is declared and how vendor reliance is handled can all affect a claim.
Part of: AI in Actuaries
In short
Many actuarial PI policies may cover civil liability arising from professional services without mentioning AI, in which case an AI-assisted error is often treated like any other model error. This is sometimes called silent AI. Some insurers may add AI-specific questions, clauses or exclusions, so check your wording at each renewal. Reliance on a vendor's model does not usually shift the client's claim away from the actuary, and recovery from the vendor depends on the contract. No regulator or IFoA statement on AI and PI has been found. Documenting AI governance against the FRC and IFoA guidance helps both underwriting and claims.
Actuarial PI is typically written on a claims-made basis and covers legal liability to third parties arising from the firm's professional services, plus defence costs, subject to the limit, excess and exclusions. Policies commonly describe the services covered, such as pensions, insurance, investment or expert work, and rely on what the firm declared in its proposal. See PI for actuaries in 2026 and PI insurance for actuaries.
On requirements, we have not found an IFoA rule requiring members to hold PI insurance. APS X5, "Compensation for Professional Shortcomings", may be relevant, so check its current text. Where an actuarial firm is FCA-authorised, the FCA's PI rules for its permission type apply. No regulator or IFoA statement on AI and PI has been found.
If a policy says nothing about AI, a negligence claim arising from AI-assisted work is usually treated like any other claim arising from the professional services, subject to the policy terms. That is generally helpful to the insured, but it is not guaranteed: other exclusions, such as those for cyber events, data or software failure, may be argued to apply depending on the facts. Some insurers may introduce affirmative AI wording or AI-specific exclusions. See affirmative AI cover versus silent AI and does PI cover AI mistakes for the wider picture.
| Situation | Usual position, subject to the wording |
|---|---|
| Machine learning model misprices a book | Usually within cover as a professional negligence claim |
| AI-generated code error across many clients | Usually within cover; aggregation clause decides limits and excesses |
| Vendor model fails, client sues the actuary | Usually within cover; recovery from vendor depends on its contract |
| Member data pasted into a public chatbot | Client claims may be covered; breach response costs are more commonly a cyber matter |
| Regulatory fine on the client | Generally not covered |
| Work not declared, such as expert witness work | Risk of dispute; declare all activities |
Illustrative scenario: a consultancy's proposal form describes its work as traditional pensions and general insurance consulting. Over two years it builds a substantial machine learning pricing practice using a vendor platform and AI-written code. A client claims that a model mispriced a book. The insurer asks why the change in activity was not mentioned at renewal and examines the firm's model governance records, which are thin. Even if cover is confirmed, the claim is harder to defend and the next renewal is more difficult. See machine learning pricing models and relying on third-party AI models.
Make a fair presentation of the risk at each renewal, and tell your insurer promptly about any circumstance that could lead to a claim. More detail: AI use and your PI: what to document.
Apex arranges PI for actuaries and actuarial consultancies; a broker can review your cover and how your policy treats AI-assisted work, and helps present your governance at renewal. See the AI in actuarial work overview.
Usually, if the claim arises from your declared professional services and no exclusion applies. Many policies may not mention AI, in which case an AI-assisted error is generally treated like any other model error, subject to the policy terms. Check for AI-specific clauses at each renewal.
Answer any questions the insurer asks accurately, and make sure significant changes in your activities, such as starting machine learning pricing or model-build work, are disclosed. A fair presentation of the risk avoids disputes when a claim is made.
Usually the client's claim is against you, and it would usually fall within actuarial PI, subject to the policy terms. Your insurer may seek recovery from the vendor, but the vendor's contract often limits that. Keep records showing you checked the model was fit for purpose.
We have not found an IFoA rule requiring members to hold PI insurance, though APS X5 on compensation for professional shortcomings should be checked. FCA-authorised actuarial firms must meet the FCA's PI rules for their permission type. Clients may also set minimum limits in contracts.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.