Guardrails

Controls that constrain what an AI system can say or do, input/output filtering, policy checks, least-privilege tool access, so it behaves safely in production.

Related terms

Need the team behind the term?

Scope your need →

All terms