During the alpha, Trust6 sends usage telemetry to the account server. This article says exactly what it contains, because transparency is the only honest relationship with confidentiality.
What is collected
Counters aggregated by day, in closed categories defined in advance. For example: how many sends happened, how many were blocked, which feature families were used, and the detection engine version. Numbers and categories, nothing else.
The telemetry envelope is closed by construction: the format simply has no field where free text, a name or a detected value could fit. That is not a moderation policy, it is a structural impossibility.
What is never collected
- The content of your conversations, prompts or answers.
- Detected or masked values (names, emails, amounts), nor their token mappings.
- The names of your vaults, clients or files.
- The content of your attached documents.
The product rule goes beyond telemetry: the account server never receives any content, even pseudonymized. It manages accounts, licences and counters.
Where to see it
Open Preferences, section Telemetry: the app shows what is collected. Data is kept encrypted locally, then transmitted periodically when the network is available.
What about my conversations, then?
They live encrypted in your vaults, on your machine, period. The only network traffic that involves them is the pseudonymized send to the AI provider you chose, the one you can inspect byte for byte.