AI Risk Assessment 2.0 - Forest Global Training Partners

Questionnaire for Test — Pax

Use case: Test assessment — AI chatbot to assist with customer support inquiries.

Security

Controls around trusted systems, operations, and security practices.

Are AI systems and their dependencies inventoried and regularly reviewed?

Is access to AI models, data, and APIs restricted and logged?

Are there procedures for detecting and responding to anomalous AI behavior?

Compliance

Legal, third-party, and governance-related controls.

Have legal/compliance teams reviewed applicable AI-related obligations for this use case?

Are third-party AI providers under contract with appropriate risk terms?

Is there a defined AI governance process or owner?

Safeguards

Safety, boundaries, dependencies, and recovery practices.

Are human oversight or intervention points defined for the AI system?

Have failure modes and dependency risks been identified?

Is there a recovery or rollback plan if the AI system misbehaves?

Culture

Human-AI relationships and delegation practices.

Are users trained on appropriate use and limitations of the AI system?

Is there clear accountability for AI-assisted decisions?

Integrity

Transparency, performance, and quality practices.

Is the AI system's performance monitored against expected outcomes?

Are limitations or uncertainties disclosed to relevant stakeholders?