County public-health leaders ask why agents need review before wide use in the city tenant. Which explanation is best?

Agents need review for data and brand risk before they are trusted across the city tenant. That explanation best answers county public-health leaders: agents can surface Microsoft 365 content and generate resident-facing language in the city’s voice, so unreviewed agents create confidentiality, accuracy, and reputational exposure. Review exists to catch those risks before wide municipal use, not as empty bureaucracy.

Claiming that agents need approval for consumption costs before they are trusted across the city tenant fails because cost approval controls spending, not the data and brand risk review required before wide city-tenant trust. Saying that admins publish agents to the app catalog before they are trusted across the city tenant fails because publishing is distribution, not a substitute for the risk review that catches confidentiality and reputational exposure before broad use. Claiming that owners save chat transcripts for record retention before they are trusted across the city tenant fails because saving transcripts is after-the-fact recordkeeping, not the pre-deployment review for data and brand risk that AB-900 requires for digital agents.

Exam caveat: Domain 3 ties approval/review to trust before wide use. On the exam, prefer answers that cite data risk and brand risk over answers that dismiss governance as delay for delay’s sake. Agents are not exempt from review because they are software.

Operational check: For each public-health agent candidate, inventory knowledge sources for sensitive data, review sample answers for brand and policy tone, confirm access is limited to intended groups, complete the approval record, and only then enable broad city-tenant trust—rechecking after major content or instruction changes.