Greetings, all. I work in an industry that deals with highly sensitive data and am considering Microsoft's AI Builder but haven't been able to find a lot of info out about how the data from files processed via AI Builder gets handled by Microsoft. We are specifically looking to train a custom document processing model.
Is this data stored anywhere on the backend when uploaded? Is it stored on shared servers or dedicated servers per customer?
Any help would be greatly appreciated.
Hi @arpost ,
Please check this documentation, it should answer your questions: https://learn.microsoft.com/ai-builder/administer
Basically, training data is persisted in Dataverse and data access control is granted by Dataverse security.
Training data is never shared externally and is never used for other purpose than training the user's model.
At predict time, no data is persisted anywhere by Microsoft.