Writing real-world programs that operate securely on untrusted hardware

Develop techniques to design and implement real-world programs that can operate correctly and securely on untrusted hardware, mitigating the security consequences of intermittent and probabilistic hardware misbehavior under current architectures.

Background

The paper highlights the gap between program properties verified at higher abstraction levels and vulnerabilities that arise at lower levels due to microarchitectural behaviors and hardware defects. This gap undermines guarantees established at the software level.

Quoting prior work, the authors emphasize that the community does not yet know how to build real-world software that can safely operate when the underlying hardware cannot be trusted, underscoring a fundamental open challenge at the software–hardware boundary.