AI incident

An AI incident is an event where an AI system causes, contributes to, or creates a credible risk of harm, failure, or rights impact.

Plan for this page

Understand one concept in plain language and connect it to nearby ideas.

You will leave with: Incidents can involve unsafe action, discrimination, privacy exposure, security compromise, misinformation, service disruption, financial loss, or a near miss. Organizations need definitions and reporting thresholds suited to their duties and context.
Time: 3–8 min
Before you begin: No prerequisite. Keep one real example in mind.
Do this now: Read the explanation, then compare the example with the common misunderstanding.

Prepared next stepStudy failure, monitoring, and recovery

AI incidentAn AI incident is an event where an AI system causes, contributes to, or creates a credible risk of harm, failure, or rights impact.

Recovery
Observability
Security

Plain explanation

Incidents can involve unsafe action, discrimination, privacy exposure, security compromise, misinformation, service disruption, financial loss, or a near miss. Organizations need definitions and reporting thresholds suited to their duties and context.

Example

An automated system repeatedly denies eligible people because a data pipeline mapped a category incorrectly, prompting suspension and remediation.

Why it matters

Incident records turn isolated failures into evidence for correction, accountability, and prevention. Good response protects affected people, preserves facts, communicates proportionately, and updates controls.

Common misunderstanding

Misconception: an incident counts only if the model crashes. Correction: a technically functioning system can still cause harmful or unjust outcomes.

Related concepts

Follow relationships, not only the alphabet

RecoveryRecovery is the ability to stop, contain, reverse, retry, compensate for, or safely continue after failure or interruption.ObservabilityObservability is the capacity to understand a system’s relevant internal state and behavior from designed records, signals, and traces.SecuritySecurity protects systems and information against unauthorized access, change, disclosure, disruption, or destruction.

Study failure, monitoring, and recovery

Sources and scope

What supports this object

OECD.AI Policy ObservatoryOECD

Provides cross-jurisdiction policy and incident context; incident definitions, reporting thresholds and verified facts vary by source and authority.

Open source

Loading page content...

…