Anthropic Tests Claude Hardware Standard for Robots and Lab Tools

Anthropic has released the Model Hardware Standard in research preview, a specification for showing AI assistants how to operate robots, lab equipment, and other hardware safely, according to The Next Web.

The standard, called MHS, stores information about how a machine works and what limits apply to its operation. For example, a robot arm maker could define speed and angle limits so Claude can use the equipment within those constraints.

Anthropic has tested MHS with Amazon Web Services, Danaher, Hugging Face, and Raspberry Pi. A waitlist is open for other groups that want to trial the framework.

The company plans to make MHS open source after the research preview, but has not provided a timeline. In one lab demo, Claude received a PDF describing an experiment and ran it on hardware that used an MHS specification.