Risk Register

AI risk register — hallucination, bias, privacy, security, and operational risks

⌘K
🏛

Risk Register

done

Track, mitigate, and review AI risks across all registered systems.

Support bot factual accuracy

hallucination · Customer Support Bot · Score: 6

Bot may generate incorrect product information for new SKUs

Mitigation: Enable citation requirements and knowledge base grounding

high
open
HR PII exposure risk

privacy · HR Onboarding App · Score: 8

Onboarding documents contain sensitive employee PII

Mitigation: Implement PII masking and data residency controls

critical
open
Provider dependency — OpenAI

provider · GPT-4 Enterprise Router · Score: 3

Single-provider dependency for critical routing layer

Mitigation: Add fallback routing to Anthropic and private LLM

high
in review
Sales forecast model drift

model_drift · Custom Analytics Pipeline · Score: 6

Model accuracy degrading on Q2 seasonal patterns

Mitigation: Retrain with Q2 data and enable drift monitoring

medium
open
Onboarding bias in document screening

bias · HR Onboarding App · Score: 3

Potential demographic bias in automated document classification

high
open
Unauthorized access to regulatory docs

security · Compliance Knowledge RAG · Score: 4

Risk of cross-tenant data leakage in RAG retrieval

Mitigation: Tenant isolation verified; quarterly penetration testing

critical
mitigated

Module: governance · AI Pass Platform