AI Companion Privacy: What Actually Happens to Your Data

People say things to AI companions they might not say to anyone else — that's part of the appeal, and part of why privacy practices matter more here than for a typical app. Most people never read a privacy policy closely enough to know what's actually happening to that data. Here's what's worth checking.

Is your conversation used to train the model?

Some platforms use conversation data to fine-tune their models by default, with an opt-out buried in settings. Others treat training-data use as an explicit opt-in choice the user makes deliberately. That distinction matters: a default-on setting means your private conversations may already be part of a training set before you've thought to check.

Who can see the raw conversation?

Even without model training, conversation logs are often visible to customer-support staff, content moderators, or third-party contractors reviewing flagged content. That's a reasonable and often necessary safety measure — moderation has to happen somewhere — but a platform should be upfront about who has access and under what conditions, rather than leaving it implicit.

What happens when you delete something

Deleting a conversation from your view and deleting it from the underlying database aren't automatically the same thing. Real deletion should remove the data from primary storage, backups on a defined schedule, and any memory system that might otherwise resurface it later. A platform that can't explain what "delete" actually does under the hood usually means it doesn't do much.

What Vantrix does differently

Vantrix's memory system stores what a character needs to remember about your relationship as structured, deletable data — not raw transcripts kept indefinitely by default. Memory editing and deletion tools exist specifically so you control what's retained, rather than only being able to delete an entire conversation as a blunt instrument.