Access & Governance
Who can see what, and the institutional guardrails around AI-generated interpretation.
Governance Configuration
- Deployment
- Private Cloud
- Role-Based Access
- Enabled
- Project Segregation
- Enabled
- Evidence Audit Trail
- Enabled
- External Model Processing
- Restricted
- Retention
- Configurable
- Human Approval
- Required for consequential updates
Role-Based Access
| Role | Data scope | Rights |
|---|---|---|
| Centre Director | All projects | View, approve, export |
| Programme Manager (PMO) | All projects | View, validate, escalate |
| Principal Investigator | Own projects | View, respond, submit evidence |
| Finance Officer | Finance & procurement | View, reconcile |
| Partnership Office | Partners & commitments | View, update engagement |
| Funding Agency Reviewer | Approved review packs | View only |
Every AI inference is labelled and versioned. Approvals, corrections and rejections are written to an immutable audit trail attached to the underlying evidence.