FIDELIS is a research architecture for a hardware-security coprocessor that models computational reality and causal execution to detect violations, adapt trust, and contain attacks or faults in embedded and autonomous systems.

It is an active research and planned demonstrator program, not a claim of universal attack detection.