Catalog record · Fraud Detection & AML
Stage as of Jul 9, 2026 · historical report, not live reconfirmation.
Universal Credit Advances fraud-review prioritisation
Primary workflow source. A named organization's workflow supported by a reviewed primary source. Stage is reported separately; this label does not establish measured outcomes or independent validation.
Last researched Oct 26, 2026. Source access and reported stage are recorded below; deployment evidence does not establish ROI.
What the research supports
The live model risk-scores benefit advance requests and refers high-risk cases for human assessment; it does not decide entitlement.
- Reported stage
- Scaled
- Technology
- Machine Learning
- Business function
- Middle Office
- Impact (legacy label)
- Not labeled
- Catalog date · unconfirmed
- Not reported
- Imported
- Oct 29, 2026
- Last researched
- Oct 26, 2026
Measures & evidence
2.5 times as effective at identifying high-risk advances versus random control, 2025–26
Original metric claim
Measurement window not disclosed
Scope, baseline & source
- Scope / denominator
- Measurement population / workflow scope not disclosed
- Baseline
- Not disclosed in this account
- Claim basis
- DWP assessment; referral effectiveness, not ROI.
Source-reported claim; methodology, baseline and independent validation may be undisclosed. Inspect the research dossier.
Original prose retained. A single comparable value or unit has not been established.
Read dossier source ↗Exact claim-to-source location has not been mapped. This link belongs to the research dossier.
One-day median additional payment delay for referred requests subsequently approved
Original metric claim
Measurement window not disclosed
Scope, baseline & source
- Scope / denominator
- Measurement population / workflow scope not disclosed
- Baseline
- Not disclosed in this account
- Claim basis
- Comparison against requests approved automatically.
Source-reported claim; methodology, baseline and independent validation may be undisclosed. Inspect the research dossier.
Original prose retained. A single comparable value or unit has not been established.
Read dossier source ↗Exact claim-to-source location has not been mapped. This link belongs to the research dossier.
Measures retain their original currency, scope and period. Capacity, adoption and forecasts are not realized cash savings.
Documented vendor relationships
No vendor relationship has been linked to this specific workflow in the curated overlay. This does not mean it was built without vendors.
See the company’s documented relationshipsDeployment evidence & limits
DWP explicitly describes a fraud-and-error ML model deployed at scale in live service. Status as of 2026-07-09.
Stage is as reported in the cited evidence. Current operating status has not been independently reconfirmed; the research date is an editorial review date.
DWP finds weaker effectiveness for non-UK nationals, couples and several age bands; protected-characteristic coverage is limited. Retrained model was still awaiting further testing before deployment at publication. Issuer judges continuation proportionate; that conclusion is not independent proof of fairness. Benefit eligibility, fraud base rates and harm from payment delays differ. Detection lift is not a financial-loss reduction estimate; subgroup and appeal monitoring must be local.
Potential FS application · inferred
Inferred FS application: Claims-fraud or disbursement-fraud review queues with a random-control sample and human decisions.
Prioritising scarce reviewers and measuring subgroup referral/outcome disparity transfers to insurance and banking fraud operations.
Original industry: Social benefits administration. Transfer limits: Benefit eligibility, fraud base rates and harm from payment delays differ. Detection lift is not a financial-loss reduction estimate; subgroup and appeal monitoring must be local.
Recorded dependencies
- Historical decisions
- Blinded human review
- Random controls
Source register
- Effectiveness Assessment of Universal Credit Advances Modelgov.uk · opens in a new tab
Department for Work and Pensions · issuer · full-text · Published 2026-07-09
Research lanes: public_scientific · Batch global-discovery-2026-10-26. Coverage and remaining gaps