Browser-style “incognito” or “private mode” toggles are appearing in a growing number of AI companion products, promising a session that supposedly leaves less of a trace. The framing borrows directly from browser privacy modes, which is worth pausing on before assuming the same guarantees apply.
What we know:
In a web browser, incognito mode has a well-documented, narrow meaning: it does not save local browsing history, cookies or form data to that device, but it does not hide your activity from your employer, school, internet provider or the websites you visit. When an AI companion app borrows the same label, the honest baseline assumption should be the same narrow scope — a local or session-level change — unless the operator’s own documentation explicitly states that a private-mode conversation is not logged, not used for model training, and not retained server-side.
What remains open:
EmberGF has not audited a specific incognito-mode implementation end to end for this piece, and “incognito” is not a standardized term in this product category the way it is in browsers — one app’s private mode could mean something meaningfully different from another’s. We cannot confirm what any single product’s toggle actually changes server-side without reading that product’s current documentation directly.
Bottom line
Before trusting an “incognito” toggle with anything sensitive, find the specific sentence in that app’s privacy policy or help documentation that defines what the mode does on the server, not just in your browser tab — and if you cannot find that sentence, assume the toggle changes less than the name implies.