A newly published patent application from Microsoft details a new system and method for using machine learning to conceal sensitive information commonly found within text.

The system makes use of a user-defined contextual privacy setting. First, input comprised of text, including sensitive information, is received by a computing device. The contextual privacy setting is updated and stored in association with the text.  Next, sensitive information is identified in the text using machine learning and a keyword, symbol, user indication, or a formatting style associated with sensitive information. In addition, the system enables users to establish varying levels of concealment relating to sensitive information, ensuring that only those with access are able to view the sensitive text.

This message will self-destruct in 5 . . . 4 . . . 3 . . . 2 . . . 1 . . .