Emergency operations staff are tempted to paste live incident rows into Copilot chat for faster queries. What is the correct guidance?
Select an answer to reveal the explanation.
Short Explanation
Production incident rows are like sealed evidence bags—do not dump them into a chat window; use scrubbed or non-prod samples instead.
Full Explanation
The correct guidance is not to paste production civic incident rows into Copilot prompts; use scrubbed or non-prod samples instead. Emergency operations staff may want faster queries, but production incident data in AI prompts exposes sensitive municipal information to the model service and conversation history.
Pasting full production incident tables into every prompt for accuracy fails by concept because accuracy does not justify production data exfiltration into chat. Disabling auditing whenever Copilot is used fails because it removes accountability precisely when AI assistance increases process risk. Storing production incident exports permanently in Copilot conversation history fails because it retains sensitive civic data in an inappropriate long-lived store.
Exam caveat: AI-assisted SQL on DP-800 still obeys data-handling rules—scrubbed/non-prod samples only. Distractors that disable auditing or retain production exports in chat history are policy failures.
Operational check: Publish an EOC standard operating procedure banning production pastes into Copilot, provide a scrubbed incident sample database for prompt use, spot-check chat histories for production data, and keep auditing enabled during AI-assisted work.