Anthropic Introduces New Standard for AI Agents to Operate Physical Devices
Anthropic has introduced a research preview of the Model Hardware Standard, a shared interface for AI agents to operate programmable devices such as microscopes, liquid handlers and robotic arms. The standard aims to reduce the time required to integrate equipment with different programming systems.
Developed with HHMI Janelia Research Campus and other partners, MHS is intended to support automated scientific and manufacturing workflows, including parameter adjustments and hardware-error recovery. Anthropic is initially sharing the standard with selected partners to develop safety evaluations before open-sourcing it.
Anthropic has introduced a research preview of the Model Hardware Standard (MHS), a shared specification designed to help AI agents safely operate physical devices in scientific research and advanced manufacturing.The company announced the initiative on August 27, saying MHS can help AI agents work with equipment such as microscopes, liquid handlers and robotic arms.
The system is designed to allow multiple devices to communicate with AI agents through a standardized interface.According to Anthropic, laboratories and manufacturing facilities can spend weeks or months integrating different types of hardware because devices often use separate programming interfaces.
MHS aims to reduce this integration work by providing a common way for programmable devices and AI agents to communicate.The standard could support more automated scientific and engineering workflows. Anthropic said AI agents could reason through steps in an experiment, adjust parameters and, in some cases, help recover from hardware errors without direct intervention.Safety remains an important part of the project.
Anthropic is initially sharing an early version of MHS with selected partners to develop safety evaluations and best practices before making the standard open source.The project was developed in collaboration with HHMI Janelia Research Campus and involves partners in areas including science, robotics, electronics and manufacturing.
The development represents another step toward connecting agentic AI with the physical world. If standardized systems become widely adopted, AI agents could eventually help researchers and engineers coordinate complex equipment and workflows more efficiently, while human oversight remains important for safety and critical decisions.
