ComplianceEnterprise security

LLM audit logging for enterprise compliance.

Every AI decision. Every user. Every model.

Compliance teams need audit trails for AI decisions. Policy Gateway logs every policy decision with structured metadata and exports to the SIEM tools security teams already use.

Schema

What gets logged.

Every request through Policy Gateway generates a structured audit record with decision metadata, user / project tags, and policy context. Prompts and outputs are not logged by default.

Fields
decision
allow, rewrite, redact, escalate, refuse
reason_code
Structured code for why the decision was made
policy_id
Which policy version was applied
policy_user
User identifier for attribution
policy_project_id
Project / app identifier for chargeback
rollout_mode
shadow, canary, or enforce
triggered_categories
Which policy rules matched
timestamp
ISO 8601 timestamp
Sample audit record
{
  "timestamp": "2026-04-18T14:23:01.445Z",
  "policy_id": "secops-redteam-v2",
  "policy_user": "analyst-42",
  "policy_project_id": "client-acme-2026",
  "decision": "redact",
  "reason_code": "PII_REDACT",
  "rollout_mode": "enforce",
  "triggered_categories": ["pii_leak"],
  "model": "abliterated-model",
  "latency_ms": 312
}
Use cases

Built for compliance teams.

Internal audits

Demonstrate AI governance controls with structured decision logs. Show reviewers which policies were enforced and why.

Incident investigation

Trace any AI decision back to the user, project, policy version, and triggered rules. Filter by reason code to find patterns.

Usage chargeback

Attribute AI usage to teams and projects. Export to finance systems for internal billing.

Policy change tracking

Every policy revision is logged. Compare decisions before and after policy changes to measure impact.

See audit logging in action.

Run a Policy Gateway simulation and view the audit trail. No credit card required.