AI Guardrails
Content filtering and safety rules for AI inputs and outputs.
⭐Guardrailsis a Enterprise Pro feature
Get Enterprise Pro →Guardrails
Filter harmful, off-topic or sensitive content before it reaches the LLM.
Active Rules
Block PII (Email, Phone, SSN)
Detects messages containing personal identifiable information.
EmailPhoneSSN
Block Toxicity / Hate Speech
Blocks messages containing harmful or hateful language.
assholebastardbitchcuntdickfuckidiotkillmoronpissprickshitslutstupidwhorehateracistslurterrorist
Off-topic Filter
Uses an LLM classifier to reject messages unrelated to your allowed topics.
Allowlist mode: Only messages related to the topics below will be allowed. Everything else is blocked by the LLM classifier.
Allowed topics:
AImachine learningsoftware developmentChimerAIAPIs
Custom Regex Blocklist
Define your own keywords to block specific content.
No terms yet
Live Test
Blocked:
Allowed: