# Methodology Note

## Purpose

The AI Governance Incident Tracker is a Kyrantis AI Policy Tools portfolio prototype for comparing public-sector AI and algorithmic governance incidents. It is designed to support policy analysis, not legal or compliance advice.

## Inclusion Standard

Incidents are included only when there is a traceable source URL and enough public information to describe the governance issue, safeguard failure, known outcome, policy lesson, and evidence-confidence rating. Cases were selected and drafted with AI assistance from public sources. These records have not undergone independent source-by-source human verification; that pass is planned.

## Source Hierarchy

1. Official inquiry, court, regulator, auditor, ombudsman, parliamentary, or government report.
2. Official agency documentation, transparency register, algorithm register, or AI inventory.
3. Reputable academic or civil-society research.
4. Reputable investigative journalism where official records are incomplete.

## Evidence Confidence

High confidence means the card is grounded in official, court, regulator, auditor, or government material. Medium confidence means it relies on strong civil-society, academic, or investigative documentation. Evidence confidence is not a legal finding.

## Claim Discipline

The tracker distinguishes official findings from reported concerns and commentary through source type, confidence labels, and case copy. Incident classifications are for policy analysis only and should not be treated as legal conclusions.

## Limitations

Public information is incomplete. Some incidents are better documented than others. Not every algorithmic incident involves AI or machine learning. The tracker is not comprehensive. Outputs and examples require qualified human review.