The Nine Trust Patterns

A trust pattern is a reusable interaction that repairs a specific trust gap.

That's doing more work than it looks like. It's not a principle ("be transparent") and it's not a feature ("add a settings page"). It's a named, portable move that closes a particular gap between what a person believes an AI system is doing and what it's actually doing — the same way a design pattern closes a particular gap between what someone wants to do and what an interface offers them.

Naming them matters for two reasons. First, a named pattern is something you can require, review, and check for; an unnamed intention is a thing you meant to do. Second — and this is the AI-specific part — a named pattern is something you can put directly in a prompt. "Make it trustworthy" produces vague, generic output. "Apply consent-at-the-moment and undo-and-override to this flow" produces work.

Nine patterns. You'll recognize most of them from products you already use — that's the point. This is a vocabulary for things good designers have been doing without a shared word for them.