What insurers need to know before they commit to an AI supplier
Before signing, ask where insurance data goes, how the tool is tested on your use case, what decisions users can challenge and how the service fails safely. Seek contractual evidence for those answers.
Review data storage, explainability, failure handling, accountability and contractual control with the teams that will use and oversee the service. Unanswered questions should remain part of the procurement decision.
The questions that matter most before signing
Use these questions to organise the supplier review:
- Where is customer, claims, and broker data stored and processed?
- What data is retained, for how long, and who can access it?
- Can the vendor explain how the model produces outputs in a way your risk, compliance, and operations teams can challenge?
- What happens when the model is wrong, unavailable, or produces inconsistent results?
- What controls exist for human review, escalation, and override?
- Has the vendor documented testing, monitoring, and model change management?
- What contractual protections cover incidents, data misuse, and service failure?
Use the answers to identify evidence gaps and decide which uses of the service the insurer is prepared to consider. A list of answers alone does not establish readiness for regulated work.
Book an AI Risk & Efficiency Audit
Why data storage should be examined first
Start the data review with the personal data, claims histories and commercially sensitive underwriting information the service would handle. Ask the vendor to document:
- the hosting region and any cross-border transfers
- whether data is used to train shared models
- how encryption works in transit and at rest
- whether sub-processors are involved
- how deletion requests and retention rules are enforced
Ask for written architecture and data flow maps, then compare them with the contract wording. Unclear or inconsistent answers leave the data-handling review incomplete.
What good model explainability looks like in practice
Ask whether your team can understand why a recommendation appeared, which inputs influenced it, the confidence or uncertainty involved and the limits of the output.
For insurers, that matters most when AI is shaping decisions around claims triage, fraud detection, customer communications, or underwriting support. Your team should ask:
- Can outputs be traced to source inputs or evidence?
- Can users challenge and override outputs?
- What testing has been done for bias, drift, and edge cases?
- What documentation exists for model updates and version changes?
If the vendor cannot provide that context, record the gap before considering the tool for high-impact insurance work.
How failure handling separates serious vendors from risky ones
Ask how the vendor handles outages, hallucinated content, low-confidence outputs, broken integrations and monitoring alerts. The review should examine how each failure is contained.
Look for fallback paths, manual review queues, incident response ownership, alerting thresholds and service-level commitments. Ask when the system withholds an answer and how that behaviour has been tested.
For a service that affects customer outcomes, agree how exceptions reach human reviewers and what operational rollback would involve.
See Managed AI Governance & Support
The commercial and governance checks insurers should not skip
Include governance and contracting in the supplier review:
- Who is accountable inside the vendor for risk, security, and model oversight?
- What audit evidence can they provide before go-live?
- How are material model changes communicated?
- What rights do you have to pause, exit, or retrieve data?
- What indemnities, liability positions, and incident notification terms are offered?
Resolve these questions while assessing suitability, before signing. Evasive answers should be recorded as unresolved procurement issues.
The safest approach in practice
Bring operations, risk, compliance, security and procurement into the diligence exercise. Use a bounded pilot to assess the evidence, agree where human review is required for the proposed use and document the response to degraded or unsafe output.
Use the findings to decide whether to consider production use, restrict the service to a low-risk pilot or stop the procurement. Keep any unresolved controls visible in that decision.
Plan an AI Implementation Project
Need a safer AI route for your firm?
If this article reflects a live decision in your practice, the useful next step is to map the workflows, confidentiality risks, supplier controls and governance gaps before tools spread informally.