Skip to content
Course · Shipping AI interfaces

When answers need disclaimers

The hallucination note is a trust feature, not a legal scar. The calm version says what to check and how, once, dismissibly — and never hides behind the fear.

Beginner1 min read (computed · recorded 3)updated 2026-09-14aihonestytrustcourse

by Motif Editors

revised 2026-09-14First publication in the bank-4 AI-interface curriculum.

Part 10 of 15 in Shipping AI interfacesAgent steps users can followLocal-first as a feature

Key takeaways
  • A disclaimer should name the action, not the terror: check the sources, run the snippet, read the diff.
  • One dismissible note beats five permanent banners — repetition teaches users to ignore it.
  • The note is part of the product's voice: a product that fears its own output is telling you something.

Name the action, not the terror

'Generated content may be wrong' is true and useless. The useful disclaimer continues: 'check the cited sources, run the snippet, read the diff before trusting it.' An action a user can take beats an anxiety they cannot. This site's note does exactly that — three concrete checks, zero fear-mongering.

One note, dismissible, repeated never

A disclaimer that appears on every answer at full volume is noise; a user who has seen it twice starts skipping it. The calm version appears once, says the checks, and can be dismissed — the dismissal is the proof the interface trusts its user. And the note is voice: a product that apologises for every answer is a product that does not stand behind its own work.

Practise the lesson

Theory sticks when you ship it. These original Motif assets put this guide's lesson to work — open one and copy it into your own page.

The component this essay works with — mounted, not pictured
Hallucination Note

Generated — verify before you ship

This answer may be wrong. Check the cited sources, run the snippet, and read the diff before trusting it.