Anthropic Previews Model Hardware Standard for AI Agent Device Control

MHS addresses device integration challenges by introducing a standardized driver that translates between operating systems and hardware devices using simple primitives like "read" and "write." It makes devices discoverable, allows agents to understand new hardware via natural language tags, and provides control mechanisms through MCP, CLI, or code files for orchestration.

Anthropic has launched a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to safely operate physical devices, initially for scientific research labs and advanced manufacturers. This standard streamlines hardware integration, reducing setup time from weeks to hours, and enables autonomous, round-the-clock experiments.

Early partners like Genentech, University of Washington, Carnegie Mellon, HHMI Janelia, and QuEra Computing have used MHS to automate lab procedures, remotely monitor instruments, accelerate dose-response experiments, speed up microscopy, and stabilize quantum computer lasers, demonstrating reduced integration time and faster iteration.