Devices, systems, and methods of detecting user identity, differentiating between users of a computerized service, detecting a possible attacker, and flagging a particular financial transaction or a particular retail transaction as being possibly-fraudulent. The methods include monitoring of user-side input-unit interactions, in general and in response to an interference introduced to user-interface elements. The monitored interactions are analyzed, and enable extraction of hardware-specific features of a computer mouse, a touchpad, a touch-screen, a keyboard, or other input unit. In some methods, detection of different mouse polling rates or different mouse DPI values, across two different usage sessions in the same financial account, enables the method to detect a possibly-fraudulent transaction.