Hashing involves taking a block of data and, through the use of a one-way operation, producing a fixed- size value that can be used for comparison with other data. It is used primarily for protecting data and allowing for rapid comparison when matching data values such as passwords. Labels involve looking for header information or other categorizations of data to determine its type and possible classifications.
Metadata involves looking at information attributes of the data, such as creator, application, type, and so on, in determining classification. Content analysis involves examining the actual data itself for its composition and classification level.

Policy drives all programs and functions in the organization; the organization should not conduct any operations that don't have a policy governing them. Penalties may or may not be an element of policy, and severity depends on the topic. Multifactor authentication and homomorphic encryption are red herrings here.

Explanation
Any application that sends data to the user will face the potential that the user could manipulate or alter the data, whether it resides in cookies, GET or POST commands, or headers, or manipulates client-side validations. If the user receives data from the application, it is crucial that the application validate and verify any data that is received back from the user.