Trust, but Verify: When AI Is Confidently Wrong

Why models fabricate specifics with total confidence, the verify list that catches it, and a personal checklist for the outputs you actually rely on.

What it covers

• Why models fabricate specifics (dates, numbers, quotes) and where it happens most.
• The verify list: every date, number, name, and quote gets checked against a source.
• "Cite it or say you don't know" as a standing instruction.
• AI summaries of meetings and threads are leads, not facts.

What you'll build

A verification checklist for the three kinds of output you use most.