Mistral Changes Default Data Training Setting to 'On' for General Users, Excluding Only Enterprise

Mistral AI has changed the default setting for using user input and output data for model training in its Vibe and API services to 'On'. General users must now explicitly opt out in the settings if they do not want their data used for training.

Scope by Service

• Vibe: General users start with training consent enabled and can opt out at any time in the settings. In contrast, Vibe Enterprise customers are excluded from training by default, and the opt-in toggle is managed via the admin panel.
• Mistral Studio and API: You can disable the 'Anonymous improvement data' toggle in the Privacy menu of the admin panel to prevent API call data from being used for training.

Opt-Out Procedure and Precautions

• Vibe: In the admin panel, select Vibe under the Manage section, then turn off the 'Allow your interactions to be used to train our models' toggle in the Privacy section. In the mobile app (iOS/Android), uncheck the 'Enable data sharing' box under Settings > Data & Account Controls.
• Independent Settings: The opt-out toggles for Vibe and the API are independent. Opting out of one does not automatically opt you out of the other, so each must be configured separately.
• Document Inclusion: Documents attached or uploaded within Vibe are considered input data, so you must opt out to ensure sensitive documents are not used for model improvement.