Government and Public Sector
Where a citizen must be able to contest the decision
Public bodies exercise powers over people who cannot take their business elsewhere. That changes what automation is legitimate. The assessment gives weight to administrative fairness, the right to an explanation, the route to challenge a decision, and the public record — obligations that exist regardless of how well a model performs.
Where AI is most often proposed
These are the use-case families the pack recognises. Appearing on this list is not an endorsement — several of them resolve to a no-AI outcome more often than their sponsors expect.
- Casework triage and prioritisation
- Eligibility assessment and benefits administration
- Correspondence handling and citizen contact
- Fraud and error detection in payments
- Document processing across licensing and registration
- Policy analysis and consultation synthesis
What actually constrains deployment here
Decisions affecting rights, entitlements or liberty
Automated determination of an entitlement, an enforcement action or an immigration outcome meets a blocking gate without an accountable human decision-maker and a genuine route to challenge.
Transparency and contestability
A citizen must be able to understand why a decision was reached and how to contest it. A system that cannot support that cannot be used for it.
Equality duties
Differential impact across protected groups must be assessed before deployment and monitored afterwards, with the assessment recorded.
Records and public accountability
Public bodies face records obligations and information rights requests. The system must be able to produce its own reasoning when asked.
Outcomes that come up more often than expected
- Hybrid human and AI, where the system prepares work and an officer decides
- Process redesign first, where the underlying policy or form is the real cause of delay
- Defer pending regulatory clarity, where obligations for the intended use are unsettled
These reflect what the framework tends to conclude given the constraints above. They are not statistics drawn from completed engagements, and no such figures are published because none have been gathered.
Content status
| Property | Value |
|---|---|
| Version | 1.0.0 |
| Maturity | Foundation |
| Validation status | Unvalidated |
| Validated by | Not validated by a named specialist |
| Functions | 18 |
| Processes | 40 |
| Use Case Families | 28 |
| Questions | 20 |
| Hard Gates | 10 |
| Risk Patterns | 12 |
| Controls | 0 |
| Alternatives | 0 |
| Kpis | 22 |
| Financial Templates | 0 |
| Implementation Patterns | 0 |
| Assessment Templates | 5 |
Internally authored by DUN-AI and not externally validated. No named specialist has reviewed this content.
A public law specialist and an information rights adviser should review this pack before its regulatory content informs a decision.