Box AI
Box says neither Box nor its model providers train on customer data sent through Box AI, and Box will only train on customer content with explicit customer approval.
Box AI Principles and support docs state that Box will not train AI models on customer content without explicit authorization, and model providers do not train on data received from Box.
No opt-out is required for the stated baseline. Administrators can enable or disable Box AI and decide whether it applies to content.
Box AI respects Box permissions, so users only interact with files they can already access. Queries retrieve relevant content under those controls.
Box says questions, answers, and document content used for a request are kept in memory only as long as needed to complete the request, and providers delete queries and results after returning a response. Browser-side history can persist while a document remains open.
The reviewed Box materials focus on training, permissions, and temporary processing rather than one general human-review statement.
Foundational reference pages for teams adopting, deploying, and operating production AI with inspectable controls.