Guardrails and AI Safety Guardrails & AI Safety 1 of 57 in this book · 1 min read Scope: runtime safeguards against unsafe input, output, and actions. Capabilities: 14 # Capability 1 Input Guardrails 2 Output Guardrails 3 Tool Guardrails 4 Prompt Injection Detection 5 Jailbreak Detection 6 PII Detection and Masking 7 Content Moderation 8 Data Leakage Prevention 9 Policy Enforcement 10 Action Risk Classification 11 Grounding Validation 12 Hallucination Detection 13 Human Approval Gates 14 Fail-closed Behavior