"An insider attack, sometimes referred to as an inside job, is defined as a crime perpetrated by, or with the help of, a person working for or trusted by the victim" . This one-sided relationship of trust makes the insider attacks particularly insidious and difficult to protect against. This article motivates the need for secure and tamper-resistant storage of the secret information that is impenetrable even by the operating system and efficient ways of meeting this need. It highlights innovative new work being developed in the context of the Trusted ILLIAC project at the University of Illinois. A progression of techniques is presented providing increasing levels of security starting from a purely software-based approach, to hardware/software partitioned and hardware-only mechanisms. This is to guard the system effectively against insiders having increasing levels of intrusive access from user-level, administrative up to even physical access to the system under threat of attack. Techniques covered include software- and hardwarebased memory randomization, hardware for a threshold cryptography enabled mechanism to allow tamper-proof key management and support the software technique. Further, we describe an Information Flow Signatures based technique to provide runtime data integrity guarantees. Reconfigurable hardware is used to ensure the secure computation of critical data. In order to enable this trusted computing hardware we explore requirements for securely initializing it under the threat of an insider attack. The unique advantage of a hardware implemented mechanism is that the secret, either the key or the code that operates on securitycritical data, cannot be revealed or modified even by the operating system.