Your privacy is a top priority when you use features like Vocal Mode in le Chat. Here’s how your voice data is handled.
Voice message privacy and default training policy
By default, your voice messages are treated as private and are not used to train our AI models.
🔑 This policy applies to all users, regardless of their le Chat subscription plan. We do not automatically use your voice message recordings or their initial transcriptions to train our general-purpose or transcription AI models.
Optional feedback for transcription improvement
You have the option to voluntarily provide feedback on the quality of a voice message transcription. This is done by using the “Thumbs up” (👍) or “Thumbs down” (👎) buttons that appear below a transcripted message card.
Example of “Thumbs up” and “Thumbs down” feedback buttons next to a transcribed voice message
🚨 If you choose to rate a transcription using these feedback buttons, that specific voice message recording and its corresponding text transcription will be sent to Mistral AI. This data is then used exclusively to help train and improve our transcription service for greater accuracy in the future.
This is an opt-in feedback mechanism. Only the voice messages you explicitly rate are used for this specific purpose. If you do not rate a transcription, your voice message is not used for training.
🔎 For more detailed information about our data handling and training policies, please visit the Trust section of this Help Center.