FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Claims management companies
CMCs can use AI to decide whether a consumer has a valid claim, but solely automated rejections may need safeguards under UK GDPR Articles 22A to 22D. A wrong screening decision can cost a consumer their claim and the firm a negligence allegation.
Part of: AI in Claims management companies
In short
AI eligibility tools score and accept or reject claims, often with no person involved. Since 5 February 2026 the DUAA 2025 rules on solely automated significant decisions require safeguards such as information, representations, human intervention and a right to contest, with extra limits on health data. Under the Consumer Duty, rejecting good claims or pursuing weak ones risks foreseeable harm, and the FCA expects outcomes testing across consumer groups. A negligent rejection may lead to a claim that PI, where held, may cover subject to the wording.
Screening is the decision at the heart of a claims business: is this consumer likely to have a valid claim, and is it worth pursuing? Some CMCs may hand that decision to software. Online questionnaires feed rules engines or machine learning models that score eligibility; some tools read uploaded documents and lender data to decide whether a claim meets the firm's criteria; others predict likely value and set the order in which files are worked. In some journeys the consumer may receive an accept or reject message with no person involved.
Section 80 of the Data (Use and Access) Act 2025 replaced UK GDPR Article 22 with new Articles 22A to 22D, fully in force from 5 February 2026 under SI 2026/82. Where a decision is based solely on automated processing and has a significant effect on the individual, Article 22C requires safeguards: giving the person information about the decision, letting them make representations, providing human intervention, and allowing them to contest it. Article 22B restricts solely automated significant decisions that use special category data, such as health data in personal injury or mis-sold protection claims.
Telling a consumer they have no claim may well be a significant decision for them, particularly if they take it at face value and do nothing further. Whether a specific screening tool is caught depends on how much meaningful human involvement there is, so firms should take advice on their own process.
Illustrative scenario. A CMC's eligibility model was trained on claims that succeeded in the past. After a change in how lenders present data, it starts rejecting a category of valid claims. Because rejection messages are automated, no one notices for months. Some consumers rely on the rejection and miss the time limit for complaining. One later instructs a solicitor and alleges the CMC's screening was negligent.
The opposite failure is as common: a lenient model accepts claims with little prospect, the firm runs them to the Ombudsman at a cost, and consumers who were told they had a good claim complain when it fails. Bias is a third risk, where a model scores some groups lower for reasons linked to age, disability or postcode. See our wider guide to AI bias and discrimination claims.
A claim that the firm negligently advised a consumer they had no claim is a professional negligence allegation, so a PI policy, where held, may respond subject to the wording and to proof of loss. Liability can be harder to establish where the firm only declined to act, which is why clear rejection wording matters. Data protection complaints, ICO action and fines are commonly handled differently, and some PI policies limit or exclude data protection liability. Many related rejections caused by one model fault may be treated as one claim under an aggregation clause. Personal injury CMCs must hold PI under CMCOB 7.4 (see AI in personal injury claims); other CMCs may not have any.
Apex arranges PI for claims management companies and can check how your policy treats automated screening decisions, data protection claims and aggregation. Start with AI in claims management.
It can be. Where a decision is based solely on automated processing and has a significant effect, Articles 22A to 22D (inserted by the DUAA 2025, in force 5 February 2026) require safeguards including information, representations, human intervention and a way to contest. Whether your tool is caught depends on the level of meaningful human involvement.
Potentially. If a consumer relied on the firm's assessment, lost the chance to recover and can show negligence, they may bring a claim. Clear rejection messages that explain the consumer can pursue matters themselves reduce the risk.
Often in personal injury and protection claims. Article 22B restricts solely automated significant decisions based on special category data such as health data, so screening that uses medical information needs particular care and advice.
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.
