AI Safety & Responsible AI
AI incident
An event where AI causes or contributes to harm, disruption or another defined unwanted result.
Example
An organization records an agent's unauthorized disclosure as an incident.
Why people use it
It gives organizations a way to record and investigate harmful AI-related events.
What you'll hear
“We should record this as an AI incident.”
What this means for you
Define reporting criteria and preserve enough evidence to investigate.
Can you control it?
No
No direct control. This describes a wider issue, concept or result rather than something you can simply switch on or off in a tool.
Common questions
- Is every incorrect AI answer an incident?
- It depends on the incident definition and consequences.
- Can an incident involve a person using the AI incorrectly?
- Yes. The event can arise from the interaction between the system, users and surrounding process.
- Why keep records of what happened?
- They help reconstruct the event and identify which changes could prevent a repeat.