FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Apex Insurance Brokers
Speak to a brokerGet a quote →

Aipedia: Insurance brokers

AI in claims notification and first notification of loss

Brokers can use AI to take and triage client claims, but a late or incomplete notification caused by the tool is still the broker's failure. Notification conditions are often strict, so an AI delay can cost a client its cover.

In short

AI now handles first notification of loss for many brokers: chatbots and forms collect details, models triage urgency and route claims, and generative tools summarise the claim for the insurer. The risks are a notification that sits in a queue past the policy deadline, a summary that leaves out a key fact, and triage that treats vulnerable customers worse. The FCA's October 2026 good practice page includes a case study on testing claims models for vulnerability bias. If a notification failure leads the insurer to decline, the client will usually look to the broker, and broker E&O cover would commonly respond, subject to its terms.

How brokers use AI at the claims stage

Claims are where clients most value their broker, and where volume can overwhelm a small team. Brokers now use web chat and voice tools to take first notification of loss out of hours, models that score claims by urgency or complexity, and generative AI to draft the notification and supporting summary for the insurer. Some systems pass claims directly into insurer portals without a person seeing them.

The broker usually acts for the client when notifying, so the timing and content of the notification are the broker's responsibility once the client has reported the loss.

The duties that apply

The FCA's October 2026 good practice page describes a firm that tested whether vulnerability-linked indicators, such as accessibility requirements, affected fraud suspicion or claim acceptance, and improved its model as a result. It also describes a firm monitoring AI-generated claims information against criteria such as completeness and faithfulness to source information. Both are directly relevant to broker claims tools.

What can go wrong

Illustrative: a commercial client reports a potential liability claim through the broker's online form on a Friday evening. The triage model classes it as "information only" because no injury is mentioned in the first answer, and it waits in a low-priority queue. The client's policy requires notification of circumstances within a set period. By the time a handler picks it up, the period has passed and the insurer relies on the notification condition. The client says it told its broker in time.

The effect of breaching a notification term depends on whether it is a condition precedent or an ordinary condition, which is why speed and accuracy matter so much.

How PI usually responds

Where the client notified its broker in time but the broker notified the insurer late or incompletely, the client's loss is usually the claim it could not recover, and it will typically bring that against the broker. Broker E&O cover meeting MIPRU 3.2.4R would commonly respond, depending on the wording. The broker's records of when the client reported, and what the AI did with the report, will often decide the outcome.

For vulnerable customers, see FG21/1 and vulnerable customers in broking.

Practical controls

See also broker chatbots and AI customer service, AI call notes and sales QA and the insurance broking AI hub.

How Apex helps

Apex arranges PI for insurance brokers. We can review how your E&O policy responds to claims handling and notification errors, including where an outsourced or AI-driven claims platform was involved.

Sources

Frequently asked

Is a broker liable if its AI system delays a client's claim notification?

It can be. Once the client has reported to the broker, the broker is usually responsible for passing the notification to the insurer promptly and accurately. An AI or platform delay does not move that duty, although the broker may have contractual rights against its supplier.

What did the FCA say about AI in claims?

Its October 2026 good practice page includes a firm that tested whether vulnerability-linked indicators affected fraud suspicion or claim acceptance and improved its model, and a firm that monitored AI-generated claims information for completeness and faithfulness to source information.

Should AI-drafted notifications be checked by a person?

For anything other than simple, standard claims, yes. A person should compare the notification with what the client actually said, check that all known facts are included and confirm the policy deadline has been met.

Does the Ombudsman look at AI in claims complaints?

The Financial Ombudsman Service said in April 2026 that it had seen very few complaints about AI so far, but it wants firms to give a clear rationale for how AI shaped an outcome. Brokers should be able to explain what their tools did with a claim.

Check your PI covers how you use AI

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 callback

Apex 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.