Enterprise Guardrails
Protect your LLM usage with content guardrails. Detect and block prompt injections, PII, secrets, and more — available for Enterprise customers.
Guardrails protect your organization by automatically detecting and blocking harmful content in LLM requests.
Built-in Protection
System rules detect common threats:
- Prompt Injection — attempts to override system instructions
- Jailbreak — attempts to bypass safety measures
- PII Detection — personal information like emails, phone numbers, SSNs
- Secrets — API keys, passwords, and credentials
- Sensitive File Types — block uploads of restricted file types
Configurable Actions
For each rule, choose how to respond:
- Block — reject the request entirely
- Redact — remove sensitive content and continue
- Warn — log the violation but allow the request
Custom Rules
Create organization-specific rules:
- Blocked Terms — prevent specific words or phrases
- Custom Regex — match patterns unique to your use case
- Topic Restrictions — block certain topics
Security Events Dashboard
Monitor all violations with a dedicated dashboard showing:
- Total violations and trends
- Breakdown by action taken (blocked, redacted, warned)
- Detailed violation logs with timestamps
Access
Guardrails are available on the Enterprise plan.
Interested? Contact us to enable Enterprise for your organization.