Home / Applied AI / 03. AI Security Safety / 00. Safety Guardrail Design Safety Guardrail Design¶ The chapters in this module, in reading order. # Chapter 00 Guardrails & Safety for Production AI — The Five-Year-Old Version 01 Why guardrails matter — fluent systems can still crash the airport 02 Input validation — the passport desk decides what is even allowed in 03 Prompt injection defense — do not let the passenger rewrite airport policy 04 PII detection and redaction — the redaction tray removes contraband before anyone travels 05 Output schema validation — the answer also needs a passport before it leaves 06 Content filtering — arrival customs checks what the model is trying to carry out 07 Refusal logic — the no-fly desk decides when not to board the request 08 Hallucination detection — arrival customs checks whether the declared facts match the manifest 09 Rate limiting and abuse control — the control tower manages traffic before the runway jams 10 Guardrail frameworks — prebuilt checkpoints versus your own airport assembly 11 Testing and red teaming — stress the checkpoints before real attackers do 12 Monitoring and incidents — the control tower must notice bypasses before the whole airport learns the hard way 13 Honest admission — what guardrails still do not solve cleanly