Shadow AI Discovery

Protect shadow AI findings, check sources, require governance review and retain investigation evidence.

Governance starter · Regulated SaaS, Private cloud

What it checks

Policies in the Shadow AI Discovery template, in the order the gateway runs them
PolicyOnChecksAction
prompt-injectionRequestDetects jailbreak and prompt-injection signalsBlock
rbacRequestEnforces role, identity, and data-sensitivity access rulesBlock
pii-detectorRequest1 custom identifier patternRedact
dlp-filterRequest3 blocked terms such as “secret” and “private key”Block
citation-verifierResponseChecks that the request context supports the citations in the outputFlag
quality-scorerResponseAccuracy ≥ 0.80; relevancy ≥ 0.78; combined score ≥ 0.85; at least 120 characters and 2 sentencesBlock
human-oversightResponseNo action is set, so it does not review responsesNone
audit-loggerRequestAdds an allow-only audit marker to the decisionRecord
Policy YAML
pack:
  id: "cfg_shadow_ai_discovery"
  name: "shadow-ai-discovery"
  version: "0.1.0"
  enabled: true
  author: "verdictan"
  description: "Shadow-AI review controls for protected findings, citation checks, a configurable human-oversight point, and traceability."
policies:
  chain:
    - "prompt-injection"
    - "rbac"
    - "pii-detector"
    - "dlp-filter"
    - "citation-verifier"
    - "quality-scorer"
    - "human-oversight"
    - "audit-logger"
policy:
  prompt-injection: {}
  rbac: {}
  pii-detector:
    action: redact
    detect_patterns:
      - "(?i)(?:national\\s*id(?:entifier)?|national\\s+insurance\\s+number|ni\\s+number|nid)\\s*[:#-]?\\s*[A-Z0-9-]{5,20}"
  dlp-filter:
    blocked_terms:
      - "secret"
      - "private key"
      - "access token"
    action: block
  citation-verifier: {}
  quality-scorer:
    min_output_chars: 120
    min_sentences: 2
    thresholds:
      min_aggregate: 0.85
      min_relevancy: 0.78
      min_accuracy: 0.80
    weights:
      relevancy: 0.5
      accuracy: 0.5
  human-oversight: {}
  audit-logger: {}