44.4 Incident handling for AI mistakes

Overview and links for this section of the guide.

The Runbook

When the AI goes rogue (infinite loop, racist output): 1. **Kill Switch:** A feature flag to disable the AI feature instantly (`ENABLE_AI=false`). 2. **Analysis:** Pull the trace ID. 3. **Fix:** Update the system prompt or blacklist the keyword.

The Apology

Have a pre-written template. "Our automated system made a mistake. We have fixed it. Here is a coupon."

Where to go next