Mistral Le Chat — privacy & data policy

Provider: Mistral AI. PromptPrivacy score: 74/100. Last verified November 01, 2026.

What happens to your data?

Mistral AI collects your account information, chat prompts, and generated responses to operate the service and provide responses. By default, your conversations and feedback may be used to train Mistral's AI models unless you disable model training in your account settings. Mistral stores your chat history until you delete the conversation or your account, and does not sell your personal data.

Does it train on your data?

Trains on your content by default, with an opt-out. Mistral AI uses user inputs, outputs, and feedback to train its large language models under legitimate interest. However, users can opt out of having their input and output data used for model training directly in their account settings.

Can you opt out?

Yes. Users can opt out of having their input and output data used for AI model training at any time via a user setting directly in their account preferences.

Data retention

Mistral Le Chat (referenced as Vibe in the policy) retains user inputs and outputs until the user deletes the conversation or deletes their account. API inputs and outputs are kept for 30 rolling days for abuse monitoring unless zero data retention is enabled. Certain administrative and compliance data (such as civil identity, contracts, and billing records) are retained longer for statutory periods ranging from 1 to 10 years.

Encryption & security

Not specified. The policy does not mention specific encryption standards (such as TLS in transit or AES at rest) or technical security controls, though it notes that service providers undergo security audits aligned with Mistral's certifications.

Is it safe for work?

No. This policy specifically excludes personal data processed in the context of business activities, operating with Mistral as a data controller. Workplace use requires Mistral's business or enterprise agreements where Mistral acts as a data processor under a Data Processing Agreement and does not train models on customer content by default.

Source policy: https://legal.mistral.ai/terms/privacy-policy?language=en-US