What is publicly confirmed?
The strongest primary source is DWP’s Algorithmic Transparency Record for CMS Predictive Analytics Compliance.
Predicts possible payment breakdown
Historic operational and transactional information is used to predict whether a compliant paying parent may become financially non-compliant.
Used for prioritisation
DWP describes the tool as supporting operational prioritisation and proactive action intended to keep cases compliant.
Not all automation is AI
HMRC income transfers, annual reviews, online reporting and benefit deductions can involve digital or rules-based automation without being machine learning.
Accuracy safeguard
This page does not claim that every CMS calculation, annual review, collection decision or enforcement action is made by AI. The confirmed machine-learning use described here concerns predictive compliance and operational prioritisation.
How CMS digital processing developed
How the confirmed machine-learning tool works
Historic CMS information
Operational and transactional data is prepared for analysis.
Pattern analysis
The machine-learning model identifies patterns linked to later non-compliance.
Prediction
A compliant case may be assessed as being at risk of future payment breakdown.
Prioritisation
The result supports operational prioritisation and possible proactive treatment.
Caseworker action
A caseworker remains involved in reviewing the case and deciding what action to take.
Primary evidence for this process
The detailed technical description, purposes, data, operation, responsibility and safeguards should be checked against DWP’s live transparency record because it may be updated.
Read the complete official recordInformation used across CMS processes
HMRC gross-income information
GOV.UK says CMS finds the paying parent’s yearly gross income from information supplied by HMRC.
Benefits and circumstances
CMS also checks benefit information and considers circumstances relevant to the statutory calculation.
Online information supplied by customers
The online account can be used to report changed circumstances, payment problems, extra payments, expenses, income, fraud or incorrect information.
Predictive operational information
DWP’s algorithm record describes the operational and transactional information used by the predictive compliance tool.
Who runs and oversees the service?
Department for Work and Pensions
CMS is a DWP service. DWP is responsible for the public record describing the predictive compliance tool.
Named teams and suppliers
The current transparency record should be used for the published organisational owner, operational responsibility, supplier involvement and technical arrangements.
National Audit Office scrutiny
The NAO investigated CMS performance, collection, enforcement, automation and service transformation.
AI, ordinary automation and digital services
Every row below now contains a direct official source. On mobile, each row becomes a full-width evidence card rather than a cut-off table.
Risks and safeguards that require scrutiny
Incorrect underlying records
Any predictive system depends on the quality of its input data. Incorrect income, payments, communications or case status could potentially affect an output.
Human reliance on a score
A human being may remain the formal decision-maker while still being influenced by an algorithmic recommendation or prioritisation.
Fairness and bias
Public transparency should show how error rates and outcomes are monitored across different groups and circumstances.
Public notice and explanation
People should be able to understand what information is processed, why it is used and how to challenge inaccurate personal information.
How to ask for records or challenge a decision
Request your personal information
DWP’s official guidance explains how to make a subject access request for personal information held about you.
DWP subject access requestChallenge a payment-amount decision
GOV.UK says a decision about the payment amount can be challenged by mandatory reconsideration, normally within 30 days of the decision letter.
Complaints and appealsContact CMS
Use the current official contact page for telephone, Relay UK, Welsh-language and accessibility options.
Official CMS contact detailsRequest non-personal transparency information
DWP publishes a dedicated route for Freedom of Information requests. FOI is for recorded public information, not your own personal case records.
Make a DWP FOI requestFocused subject access request wording
Please provide all personal information held about me relating to any predictive analytics, machine-learning, algorithmic scoring, compliance prediction or risk classification applied to my Child Maintenance Service case.
Please include, where held:
- each score, category, prediction and date generated;
- the information and explanation shown to caseworkers;
- any recommended or suggested action;
- records showing who viewed or acted upon the output;
- caseworker comments, acceptance, rejection or correction;
- resulting calls, messages, work items or case actions;
- audit records relating to the use of the output; and
- the applicable retention period and privacy information.
Questions DWP should answer publicly
When was the first real CMS case scored?
Request the first pilot date, first production date and first date on which an output was shown to a caseworker.
What complete data fields and engineered features are used?
Request a full feature dictionary, including definitions, transformations, exclusions and version history.
How many false positives and false negatives occur?
Request monthly confusion-matrix figures and model performance broken down by model version and relevant equality groups.
How often do caseworkers disagree with the output?
Request figures for accepted, rejected, overridden and corrected predictions, together with resulting actions.
Are affected customers told that a predictive score was used?
Request the exact customer notices, scripts, letters and online privacy wording used for predictive compliance processing.
What contracts and costs support the tool?
Request procurement records, statements of work, supplier costs, hosting costs, assurance work and ongoing support costs.
Primary official sources
Accuracy and legal notice
This is an independent public-interest information page created for Right First Time UK. It is not operated or endorsed by DWP, CMS, HMRC or the National Audit Office.
Official pages can change. Visitors should use the direct evidence buttons to check the latest wording. Analysis and transparency questions are clearly separated from confirmed official statements.
This page provides general public information and is not legal, financial or data-protection advice.