The shift from Industry 4.0 to 5.0 has created highly interconnected manufacturing systems, but this expansion increases cybersecurity risks. Existing software- and hardware-based solutions are mostly reactive, which is insufficient for real-time, safety-critical environments that demand proactive and resilient protection. This paper addresses this critical gap by proposing a taxonomy of hardware-based approaches for detecting and mitigating software cybersecurity violations, grounded in cybersecure-by-design principles. The taxonomy is developed through a conceptual synthesis and comparative analysis of state-of-the-art literature, structured across dimensions such as execution layer, security function, and design paradigm. It extends current detection-focused models by classifying approaches that embed security at the hardware level to prevent or contain threats autonomously. In doing so, it introduces a novel perspective by highlighting hardware not only as a trusted anchor but as an active defender of software integrity, thereby contributing a systematic foundation for cross-layer resilience in smart manufacturing systems. The taxonomy’s relevance is demonstrated through its application to smart manufacturing systems, highlighting its value for system designers, cybersecurity engineers, and policymakers. This contribution offers a structured foundation for developing next-generation industrial systems that are secure not only by detection, but by architecture.