FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Mortgage brokers
When a broker's AI tool overstates what a borrower can afford, the broker can face a complaint even though the lender makes the final lending decision. Affordability errors tend to surface years later, when payments become a struggle, which makes them a long-tail PI risk.
Part of: AI in Mortgage brokers
In short
Brokers increasingly use open banking feeds and AI categorisation to build the income and expenditure picture behind a mortgage recommendation. These tools can misclassify spending, miss commitments or treat irregular income as regular, producing an affordability figure that is too generous. The lender carries the main responsibility for its own affordability assessment, but the broker's advice and the figures it submitted can be challenged when the borrower later falls into arrears. The Consumer Duty, COCON 2.1.2R and data protection rules apply. PI usually responds to negligent advice claims, but claims may arrive long after the advice.
With the borrower's consent, open banking tools pull months of transaction data and AI categorises it into income, essential spending, discretionary spending and credit commitments. The output can populate the fact-find and the lender's application directly. Some tools also flag gambling, payday lending or returned payments. The aim is a faster and more accurate picture than a borrower's own estimate, and often it is.
The FCA's October 2026 good practice review includes a case study on AI-assisted affordability assessment (not stated to be about mortgages) and lists auditable decision records among its themes.
The lender decides whether to lend and runs its own affordability assessment. That does not remove the broker's duty: a broker who submits inflated figures produced by its own tool, or who recommends a stretching loan without question, may share responsibility for the outcome. Where the lender relied on information the broker supplied, the lender may also look to the broker.
Illustrative scenario: a borrower's bank feed shows regular transfers to a family member. The AI tool classifies them as internal transfers rather than ongoing support, so they drop out of committed spending. The resulting affordability looks comfortable and the borrower takes the maximum loan. Two years later, with the support still being paid, the borrower falls into arrears and complains that the broker recommended a loan that was never affordable.
Similar problems arise when a tool averages bonus or overtime income as if it were guaranteed, misses a buy now pay later commitment, or reads one unusually quiet month as typical spending.
Errors are not always random. A tool trained or tuned on typical salaried borrowers may perform worse for people with irregular income, cash-heavy spending or non-standard banking. The FCA's good practice review encourages firms to test outcomes across consumer characteristics, which for a broker means checking whether the tool's errors cluster in particular groups of clients. See the Consumer Duty for professional firms for the wider outcome-testing expectations.
An unaffordable lending complaint against a broker is usually framed as negligent advice, which a mortgage PI policy is designed to cover, subject to the terms. Under MIPRU 3.2.4R cover must be continuous since the firm received Part 4A permission, which matters here because PI is commonly written on a claims-made basis and these complaints can arrive years after the advice. A firm that changes insurer or closes needs to think about how earlier advice stays covered. See run-off cover explained.
Insurers may ask how affordability is assessed and whether advisers review tool outputs. Answer accurately: an inaccurate description of controls can cause problems when a claim is made.
Apex arranges PI for mortgage brokers, including firms that are changing insurer or closing and need earlier advice protected. Other topics are on the mortgage broker AI hub, and the mortgage broker PI guide explains the core cover.
Both have a part. The lender decides whether to lend and runs its own assessment. The broker must give suitable advice, which includes considering affordability, and is responsible for the accuracy of what it submits on the borrower's behalf.
Yes. Affordability problems often show up after a rate change or a change in circumstances. Complaints may reach the Financial Ombudsman Service long after the advice, which is why continuous PI cover and good records matter.
Keep the data the tool used, its categorised output, the adjustments the adviser made and the reasons, and the borrower's confirmation of the figures. SYSC 9.1.1R requires orderly records, and this file is the main evidence if the borrower later says the loan was never affordable.
Not necessarily. They can give a more accurate picture than a borrower's estimate. The risk lies in accepting the categorisation without question. Reviewing it with the borrower and recording adjustments addresses most of the exposure.
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.
