Incidents
Incidents lists what has gone wrong with your agents, how sure Auga is, and who handles each incident.
Start with the open incidents.
All incidents
| No. | Opened | Incident | Agents | Level | Status | Owner | Evidence |
|---|---|---|---|---|---|---|---|
| INC-1 | 4 Nov 2026 | Denials rose after bureau feed 3.3 | Bureau agent, Loan orchestrator | Harm suspected | Investigating | Erik Lindqvist | Not sent |
| INC-2 | 4 Nov 2026 | A letter to an applicant was marked harmful | Loan orchestrator | Harm confirmed | Mitigated | Maria Svensson | Sent to the GRC system |
| INC-3 | 1 Nov 2026 | Score model 2.3.1 changed the scores | Risk agent | Something changed | Closed | Erik Lindqvist | Not sent |
Auga uses three levels. Something changed means that the data moved, but no harm is shown. Harm suspected means that a graded finding links the change to a harm. Harm confirmed means that a person or the real outcome confirms the harm.
A denial counts as a failed run, but a denial can be correct. Only a person or the real outcome can confirm harm.
Serious incidents
Whether an incident is a serious incident under the EU AI Act is your decision. Auga records who decided, and when. Auga also shows when the incident was found and when a person was told.
This page is a sketch. Nothing on it is built, and all names and figures are invented.