How private should a conversational AI companion be?
Short answer
Privacy is not only about encryption or a policy page. It is also about what the product chooses to collect, what becomes memory, and whether you can inspect and remove what is kept.
Collection is a product decision
A companion can know you better over time without trying to observe everything. The important distinction is between context that is necessary for the experience and passive surveillance that exists because it is technically possible.
Good privacy design reduces unnecessary collection instead of asking policy language to justify it later.
Control should be visible
Xiara keeps long-term memory explicit and user-manageable. It does not treat every conversation detail or every song you play as something that should become durable memory.
The practical test is simple: you should be able to understand what the product knows and change what it keeps.
Related answers