Policy Engine

📄 dlp-policy.yaml

name: "pii-protection"
version: "1.2.0"

rules:
  - match: "ssn"
    action: "redact"
    replacement: "[SSN]"
    
  - match: "credit_card"
    action: "block"
    alert: true

  - match: "email"
    action: "redact"
    log: true
                

📄 access-control.yaml

name: "model-access"
version: "2.0.0"

rules:
  - model: "gpt-4"
    allowed_groups:
      - "engineering"
      - "product"
    max_tokens: 8000
    
  - model: "claude-3"
    allowed_groups:
      - "all"
    rate_limit: "100/hour"
                

Policy capabilities

Everything you need to manage AI security at scale

📝

YAML/JSON Policies

Human-readable policy definitions that live in your repo. No proprietary DSLs or GUIs required.

🔄

Version Control

Git-native workflow. Track changes, review PRs, and roll back to any previous policy version.

🧪

Policy Testing

Test policies in staging before production. Dry-run mode shows what would be blocked.

📊

Audit Trail

Every policy change logged with who, what, when. Export to SIEM for compliance reporting.

🏷️

Tag-Based Rules

Apply policies by team, project, environment, or custom tags. Granular control without complexity.

⚡

Real-Time Enforcement

Policy changes take effect immediately. No restart, no deployment, no downtime.

GitOps workflow

Manage AI security policies like you manage code

Write

Define policies in YAML files in your repository

Review

Submit PR, get peer review, run policy tests

Merge

Approve and merge to main branch

Deploy

BladeRun auto-syncs and enforces instantly