Disclosure and Setting Expectations

Honest labels and controls

Users should know when content is AI-generated, what the feature can and cannot do, and how to check or correct it. Show sources for factual answers, make editing easy, and let users turn the feature off where appropriate. Avoid wording that implies certainty or human authorship. Clear expectations reduce over-trust and complaints, and some regulations and platform policies require disclosure of AI-generated content or automated decisions.

The human side of a launch

Clear communication, support readiness and an AI-aware incident process complete the plan.

In-product disclosure elements

Small touches that set expectations.

- label: "AI-generated summary - check important details"
- sources: links to the messages / documents used
- controls: edit, regenerate, "not helpful" with a reason, turn off
- first-use tip: what the feature does well and its known limits
- help article: how it works, what data it uses, how to opt out

Make feedback one click

The easier it is to say "not helpful", the more signal you get about failures.

त्वरित जाँच: Why label AI-generated content in the product?

• To hide how it works
• So users know to check it and can calibrate their trust
• Labels improve model accuracy
• It removes the need for review