PART 4Prevention
Slides 18–25 · 6 mitigation categories
Slide 18 · Prevention Overview
Six mitigation categories. Each addresses a different link.
No single control stops all four types. You need layers.
📂M1 — Retrieval-Augmented Generation (RAG)
Ground the model’s answers in verified external sources rather than relying on memory.
🔗M2 — Citation Enforcement
Require verifiable source links for every factual claim, and validate those citations automatically.
👤M3 — Human Review Gates
Require human review for high-stakes outputs before they reach regulated contexts.
🎤M4 — Uncertainty Disclosure
Build models and prompts that express uncertainty and surface that signal to users.
📢M5 — User Education
Teach users that verification is their responsibility, not the model’s.
🔍M6 — Output Validation Pipelines
Build automated checks that verify factual claims before returning responses to users.
No single control stops all four misinformation types. RAG (M1) and human review (M3) cover the most ground. Citation enforcement (M2) and output validation (M6) add the next layer. User education (M5) closes the behavioral gap that technology alone cannot fix.