Skip to content
AI SYSTEM — Artificial Intelligence
deen

Guardrails

Also: Leitplanken · Schutzmechanismen · Content Filter

Safeguards that check an AI system’s inputs and outputs and prevent unwanted content or actions.

Guardrails filter confidential data, offensive content or topics outside the intended use, check answers for grounding in sources or block risky tool calls, for example.

They are used in addition to the model’s built-in safety rules and can be tailored to your own requirements.

General information, not legal or professional advice.