Model Hardware Standard (MHS) is a provisional or research-preview interface concept for connecting AI agents to hardware such as robots, lab devices, or industrial tools. It is best understood as a bridge between model output and machine action, not as a settled EU standard. In practice, the interface defines how an AI can request movement, trigger tools, or change device state.
MHS matters in cyber security because the interface becomes part of the attack surface. If an attacker can tamper with prompts, permissions, middleware, drivers, or calibration data, they may turn a normal model decision into an unsafe physical action. Defenders therefore need allowlists, authentication, logging, human override, and emergency stop paths. MHS also affects compliance: any drift in hardware descriptions or access rules can weaken a safety case even when the model itself has not changed.