Studio Risk Register
The Risk Register tracks AI-specific risks at the Studio level — model hallucinations, bias concerns, dependency risks, and compliance exposure — with mitigation owners and review dates.
AI-specific risk categories
Hallucination Risk — Risk that agent output contains plausible but incorrect information accepted without adequate review.
Bias Risk — Risk that agent output reflects training data biases affecting fairness, accessibility, or protected characteristics.
Dependency Risk — Risk from reliance on third-party AI providers (price changes, API deprecation, data residency changes).
Scope Creep Risk — Risk that agents autonomously expand beyond their intended task boundaries.
Compliance Risk — Regulatory gaps between current Studio governance and applicable standards (EU AI Act, ISO 42001, etc.).
Key Person Risk — Risk if the Studio's main operator is unavailable — who can review runs and approve releases in their absence?
Auto-identified risks
The Studio automatically flags risks when: DPIA sections are incomplete, a model with high hallucination rate is used for a high-trust run, an agent runs beyond its allocated time or cost cap, or a Verify check has been excepted more than twice.
Review cycle
Set a review date for each risk. Overdue risks appear in the Studio health indicator and are surfaced in monthly governance reports.