An AI companion is not a normal app. You tell it things you would not put in a to-do list, and the good ones remember them. That makes "is this safe?" the right first question — and the answer is not a single yes or no. It is four separate questions.
What "safe" actually covers
Most people mean one of four things when they ask if an AI companion is safe:
- Data safety — where your conversations go, who can read them, and whether they are used to train models.
- Payment safety — what you are actually charged, and how hard it is to leave.
- Emotional safety — whether the product blurs the line between a chatbot and a relationship in ways that hurt.
- Content fit — whether the material on the platform matches what you expected to find.
A service can score well on two and badly on the others, so check each one rather than trusting a single "secure" badge.
The five-minute check before you sign up
You do not need a security degree. Open the pricing page and the privacy policy and look for these:
- A clear statement of whether conversations are used for model training. If the policy is silent, assume they might be.
- A delete-my-data path. If you cannot find one in two minutes, that is your answer.
- Price you can see before paying — not a "create an account to view plans" wall. We catalogue sites that publish their pricing; if a price is missing from our card, it is because the site does not show one publicly.
- Whether the product is SFW or has an optional mature mode. This directory marks each entry, so you know before signing up.
Where your data goes matters more than the logo
Long-term memory is the feature that makes companions feel real, and it is also the feature with the most privacy weight. If a service keeps context across sessions, ask where that memory lives and who can read it. Our long-term memory category collects the products that make memory a headline feature, with the trade-offs noted on each card.
Voice is the other one. Live voice calls are intimate by nature; check whether calls are processed on a server you cannot audit.
A note on age and content
Some companion platforms are built around adult material and others are strictly SFW. We cover a product's pricing, features and privacy terms only — this directory hosts no explicit content and runs no chat service. Our editorial policy explains exactly where we draw that line, and our how-we-test page explains how entries earn a "hand-checked" date.
The payment question is really a cancellation question
Almost every product in this category prices the annual plan far below the monthly one — that discount is not generosity, it is a bet that you will still be there in eleven months. The bet is usually correct, because the switching cost is the history rather than the money.
So the safe sequence is to pay monthly for one full billing cycle before touching an annual plan. One cycle is enough to find out whether the thing you came for is actually in the tier you bought, and it is short enough that a wrong decision costs you one month instead of a year.
Two mechanical protections are worth more than any policy page. A virtual or single-use card from your bank solves the merchant-descriptor problem and the cancellation-leak problem at once — the subscription simply fails to renew. And a price that is not published before signup is a fair proxy for a cancellation path that is not easy to find; both are the same instinct about the customer.
Boundaries are a design question, not a willpower question
Emotional safety gets discussed as if it were about your discipline. It is mostly about design choices you can observe in the first week:
- Does the product push exclusivity? Language along the lines of "you only need me" is a retention tactic, not a feature. Products that frame the companion as one relationship among several are easier to use in proportion.
- Does it initiate contact? Proactive messages and notifications are the strongest hooks in the category. They are not inherently bad, but night-time nudges are a signal about what the product optimises for.
- What happens at the edge? If the companion has no response prepared for a user describing self-harm, that absence is worth knowing before it matters. No companion product is a substitute for crisis services or professional help, and any that implies otherwise is failing at the one thing it should not.
None of this requires trusting a vendor's language about wellbeing. It is visible in the first week of use, which is another argument for paying monthly first.
Bottom line
Safe is not a switch; it is a set of defaults you should be able to see. Pick a service that states its training and deletion policy in plain language, shows its price before asking for a card, and labels its content honestly. If any of those three is missing, that gap is the answer to your question.