AI Governance and vCISO Readiness Diagnostic

A leadership-oriented diagnostic for security leaders, vCISOs, advisors, consultants, and GRC teams translating AI risk into controls, policy, evidence, and buyer-facing trust posture.

60 q · 75m

Question 1 of 602%

secure architecture design

Select the five controls that belong in a secure architecture for a multi-tenant RAG agent with CRM access.

The agent answers customer-support questions using tenant documents and can draft CRM updates for support agents to approve.

A. Security evals for cross-tenant leakage, prompt injection.B. Tenant and user ACL filtering before document chunks enter model context.C. Provider data-processing review covering training, retention, subprocessors, and region.D. A system prompt saying the agent should be secure, with no other controls.E. A single shared admin token for all CRM updates to simplify implementation.F. Prompt, retrieval, model, tool-call, approval, and output logs with privacy controls.G. Scoped CRM permissions that allow draft proposals but require human approval for writes.