CossoSecurity and data
Your voice stays on your computer.
Here is what leaves it.
Speech recognition runs on your computer. With AI editing on, the text goes to our server and a language model, and neither keeps it. With AI editing off, dictation works fully offline.
What goes where
Each kind of data
and where it's kept.
DataWhere it goesKept
Your voiceNowhere. Recognized on your computer by Whisper.Never uploaded
Dictated text, AI editing offStays on your computer.Last 100 in your History
Dictated text, AI editing onSent to Cosso's server, which passes it to the language model and returns the edit.Not stored
Usage countsCosso's server, for free and paid limits.Characters and timings only, no text
Personal dictionaryYour computer.Until you delete it
Team dictionaryCosso's server, so every member gets it.Until an admin removes it
AccountCosso's server: your email, for signing in.While the account exists
In practice
Common questions
from IT and compliance teams.
- Can we use it without sending anything anywhere?
- Yes. Turn AI editing off and Cosso never contacts our server while you dictate. Recognition, filler-word removal and your personal dictionary all work offline.
- Which AI providers see the text?
- Requests are routed only to model providers whose policy is not to store or train on the data. The text is passed through for the edit and not kept by us.
- Do you keep recordings for training?
- No. Audio is never uploaded, so there is nothing to keep.
- What does the server log?
- Request metadata such as timings, sizes and errors. Not the text you dictate or get back.
- Who can see a team's dictionary?
- Members of that team. Owners and admins can add and remove words.
- How do we report a security issue?
- Email the details to [email protected]
Need a security review?
Write to us for a security questionnaire, data processing terms or a call with your IT team. [email protected]