Mistral now trains on user input by default, except on enterprise tier

Mistral now trains its models on user input by default, except for enterprise customers. This change applies to conversations, documents, and other user-provided content that users may share with the platform. Users have the right to opt out of this data processing at any time, but the opt-out process varies depending on the service or platform being used.

In the Vibe version, users retain control over this processing and can opt out by accessing the Privacy section and disabling the toggle labelled "Allow your interactions to be used to train our models." This ensures that user input and output data will no longer be used to improve Mistral's models.

Enterprise customers, on the other hand, are opted out of training by default. The opt-in toggle for this feature is managed at the admin level. To opt out of training in the enterprise tier, customers can disable the toggle labelled "Allow your interactions to be used to train our models" in the Admin panel under the Privacy section.

For users accessing Mistral's services on iOS and Android devices, the opt-out process involves navigating to Data & Account Controls in the Account section and deselecting the "Enable data sharing" checkbox. This action opts the user out of Mistral's training program, ensuring that their input and output data will not be used for model training purposes.

It is important to note that customers retain full control over this processing and can opt out at any time. The opt-out toggles for Mistral Studio and related API services are separate from those of the Vibe and API services. Therefore, users must configure each toggle individually to ensure that their data is not used for model training.