Overview
Flag or block unsafe prompts/outputs and export audits for review.
Problem
Manual moderation does not scale and lacks audit trails.
Solution
Enable pre/post moderation using OpenAI moderation (if configured); choose block/warn/allow and record to AiAudit.
How it works
Set policy mode, run the assistant normally, and review flagged entries via CSV export with SAML-gated access.
Who is this for
Compliance
Security
Trust & Safety
Expected outcomes
- Reduced policy violations with clear logs
- Faster review cycles via exports
Key metrics
Unreviewed flagged items (backlog)
Baseline
120 count
Target
5 count
Mean review time
Baseline
48 hours
Target
4 hours
Gallery
Downloads & templates
Case studies
Marketplace reduces abusive content
Warn mode cut repeats by 68% with zero false blocks reported.
Marketplace Mid-market EU
Security impact
- Prompts/outputs & moderation categories · PII: none
Compliance
- SOC2
- GDPR (auditable processing)
Availability & next steps
Free
Pro
Enterprise