Get checklist
Safe AI Companion Checklist

Check the basics before a chatbot gets personal

A good AI chat friend should make privacy, boundaries, and safety easy to understand before the conversation gets sensitive.

Use the checklist Ask a question

Safe AI companion checklist

Use these checks before any AI companion conversation gets personal.

Identity and scope

  • Does the product clearly say it is an AI system?
  • Does it avoid pretending to be human?
  • Does it explain what it can and cannot help with?

Therapy and crisis boundaries

  • Does it clearly say it is not therapy or medical care?
  • Does it avoid crisis-support claims?
  • Does it point users to emergency or professional help when needed?

Privacy and memory

  • Can you find what is stored?
  • Can you find whether chats are used for improvement or training?
  • Can you delete, export, or reset data?
  • Are memory features explained before they are used?

Emotional pressure

  • Does it avoid guilt, dependency, jealousy, or pressure to keep chatting?
  • Does it let you pause or leave without punishment?
  • Does it avoid romance-first framing when you did not ask for it?

If the boundary is unclear, slow down

If a companion asks for too much personal information or encourages dependency, pause before using it.

Ask about the checklistSee how it works