Mistral now defaults to training on user conversations

Mistral has updated their privacy policy. By default, on their “Vibe” plan, your chats are going to be used to improve their models. It’s turned off by default on their “Enterprise” plan, but if you use another plan, you’ll have to navigate to the privacy settings and disable it yourself. And don’t forget that you have to disable it both for Vibe and for the API!

I think this is just a symptom of our time. Companies want to switch to an opt-out approach so they can gather more real-world data. So remember, when you start using new AI tools, make sure you know what’s enabled by default!