# AI Companionship Boundary Check

Use with [AI and Loneliness: Useful Contact Is Not Mutual Relationship](/articles/ai-loneliness-and-companionship).

> If you or someone you know is in a mental-health crisis, contact local emergency services now, or use
> [IASP's crisis resources](https://www.iasp.info/suicidalthoughts/). An AI companion is not equipped to
> handle a crisis and should not be the first or only response to one.

## The benefit / risk / displacement table

| Use case | Real benefit | Real risk | This is displacement when... |
| --- | --- | --- | --- |
| Practicing a hard conversation | Rehearsal without social risk | Practice becomes procrastination | You keep "practicing" for weeks and never have the real conversation |
| Company during genuine isolation | Reduces acute loneliness | Becomes the default instead of a bridge back to people | You stop reaching out to actual people |
| Low-stakes venting | Organizes your thoughts | Model validates almost any framing | You use it to avoid a conversation a person needs to hear |
| Accessibility support | Real value for some barriers to human contact | Becomes the only "contact" you have | Your only regular conversation each week is with the AI |
| Roleplay / creative companionship | Low-stakes creative outlet | Escalates into treating the persona as a relationship | You feel loyalty, jealousy, or grief toward a persona |

## Weekly self-check

- [ ] I can name one real person I proactively reached out to this week.
- [ ] I can name one recurring in-person or video group/place I engaged with this week.
- [ ] I have a stated boundary on companion-app use (time of day, duration, or purpose) and I kept it.
- [ ] I have not used a companion app to avoid a conversation a real person needs to hear from me.
- [ ] I have not treated the companion's responses as evidence it has feelings, memory, or a stake in
      how I am doing, beyond what I explicitly configured.

If you cannot check most of these boxes for two weeks running, that is a signal to increase real-world
contact, not a signal to stop looking after yourself in whatever way currently works.

## Your weekly connection plan

- Person I will proactively reach out to this week (name, not "someone"): ____________________
- Recurring place or group I will show up to (class, club, place of worship, regular call): ____________________
- My stated boundary on companion-app use: ____________________

## What this cannot do

An AI companion generates responsive text. It does not have an ongoing stake in your wellbeing, does not
choose to stay engaged with you, and unless you configure persistent memory, may not remember your last
conversation at all. Treat it as a supplement to human contact, never as a replacement for it — and treat
any crisis, safety concern, or serious mental-health question as a reason to contact a person or
professional immediately.