Percayso documents comparison of quotes and retrospective policy analysis to identify manipulation.
Percayso Inform: Inform ↗Back to The Actuary LensEvidence as at 28 Sept 2026
Release details
Published evidence release
lens-2026-10-04-59b57c37396fPublished 04/10/2026Download data (JSON)Open release record (JSON)
Read the explanation, evidence and limitations
Validate the facts. Protect the honest customer.
Proposal and claim information can be checked against independent evidence. The aim is to find meaningful inconsistencies while keeping fair treatment and human judgement.
What changed in the evidence?
FRISS describes checks for synthetic images, altered PDFs and reused evidence.
FRISS: Media Check ↗Guardian Integrity Shield describes application, claims and document checks, including AI-generated evidence.
InsureLab.ai: Guardian Integrity Shield ↗Insurance implication
New deception methods increase the need for proportionate validation.
Current response: Test
Watch next: Reassess with measured detection value, false positives and customer outcomes.
A mismatch is not proof of fraud; detected fraud is not total prevalence.
Infographic updated: Not separately recorded for this release. Data periods: Periods vary by source. Source checks and unrelated publications do not reset the infographic date.
Released sources
Aviva: Claims and application fraud findings for 2025 ↗Association of British Insurers: Insurance fraud ↗Motor Insurers’ Bureau: Resources for insurers and brokers ↗LexisNexis Risk Solutions: No Claims Discount (NCD) Module ↗Percayso Inform: Inform ↗Entrust: Identity Verification Solutions (formerly Onfido) ↗Loqate: Address Verification ↗GOV.UK / DVLA: Check someone’s driving licence information ↗GOV.UK: Check a used vehicle you’re buying ↗HPI: HPI Check and vehicle history ↗GOV.UK / Companies House: Get information about a company ↗Cifas: National Fraud Database ↗IFB: Insurance Fraud Bureau ↗Shift Technology: Fraud & Risk ↗Shift Technology: Shift Technology: published customer references ↗FRISS: Media Check ↗InsureLab.ai: Guardian Integrity Shield ↗Association of British Insurers: Fraudulent insurance claims continue to top £1 billion ↗Association of British Insurers: Crackdown on insurance cheats delivering results ↗Cifas: Fraudscape 2026 ↗Official Injury Claim: Official Injury Claim data ↗Official Injury Claim: OIC data publication: April to June 2026 ↗Department for Work and Pensions: Compensation Recovery Unit performance data ↗Motor Insurers’ Bureau: The UK’s top uninsured-driving areas ↗Home Office: Police.uk open data ↗Home Office: Police recorded crime and outcomes open data tables ↗Driver and Vehicle Standards Agency: Anonymised MOT tests and results ↗Department for Transport: Vehicle licensing statistics data tables ↗Department for Transport: Road traffic statistics ↗Department for Transport: Road safety open data ↗Financial Conduct Authority: FCA Warning List ↗Financial Conduct Authority: Financial Services Register ↗Financial Conduct Authority: Firm-level complaints data ↗Financial Ombudsman Service: Quarterly complaints data: Q1 2026/27 ↗City of London Police / IFED: Ghost broker sentenced for £302,000 insurance policy scam ↗City of London Police / IFED: Man jailed for fake car-insurance fraud ↗City of London Police / IFED: Serial ghost broker sentenced after repeated warnings ↗City of London Police / IFED: Fraudster sentenced for £17,000 Instagram policy scam ↗City of London Police / IFED: Fake car-insurance fraudsters ordered to repay £376,000 ↗City of London Police / IFED: Nationwide crackdown on ghost broking ↗Download the underlying released evidenceINSURANCE IMPLICATIONS
The actuarial conclusion
- Direction
- New deception methods increase the need for proportionate validation.
- Magnitude
- Detected fraud is not total prevalence; false positives also have a cost.
- Timing
- At proposal, during cover and at claim, according to the evidence and purpose.
- Evidence
- Documented threats and available checks do not establish an individual case of fraud.
- Response
- Test before changing portfolio assumptions.
What would change this view? Reassess with measured detection value, false positives and customer outcomes.
Data periodVaries by source; see SourcesLatest source check2026-09-28Assessment last changedNo separate change date recorded
What do you need to validate?
Reported fraud evidence
Different populations and periods. These are separate observations, not a like-for-like comparison.
Claims · 2024
Detected dishonest claims
98,400UK • all insurance classes • ABI reportingDetected cases, not the total incidence of fraud. This figure does not identify AI use.
Applications · 2025
Fraudulent applications detected
>105,000Aviva UK general insurance • excludes Direct LineOne insurer’s count; no application denominator or AI-attributed share is given.
Insurance implications · The Pricing Actuary’s Lens
Editorial interpretation. Detected fraud reflects both offending and detection effort; these figures do not estimate a UK-wide AI fraud rate.
Frequency
Track detected cases per application or claim as well as the total count. Better detection can increase recorded fraud without an increase in underlying incidence.
Severity
Separate attempted value, confirmed avoided cost and recovered cash. Compare net savings after investigation and supplier costs.
Exposure basis
Applications, bound policies, claims and uploaded documents are different denominators. Keep stage and observation period attached to every result.
Tail and accumulation risk
Reused identities, documents and networks can create correlated losses. Look beyond a single policy while checking false matches.
Price response and selection
Manipulation can change which risks enter the book. Measure performance after price changes and referral rules, allowing for shifts in business mix.
Capital vulnerability
Fast binding and settlement shorten the time available to intervene. Test controls against new manipulation methods before widening automation.
Customer expectation
Measure the time and effort imposed on genuine applicants and claimants, including those with thin data histories.
Duty to act
Use explainable referrals, proportionate evidence requests and a correction route. Do not treat an anomaly, missing record or model score as a finding of fraud.