Anthropic launches standard for AI agents to run lab and factory equipment

Anthropic on Thursday unveiled a research preview of the Model Hardware Standard, a framework designed to let AI agents operate physical devices in scientific laboratories and manufacturing facilities, marking the company's first move into what the industry calls "physical AI."

A Universal Interface for Machines

The MHS enables AI agents to control instruments such as microscopes, liquid handlers, robotic arms, and quantum computing equipment simultaneously, executing autonomous workflows around the clock with minimal human oversight. Anthropic developed the standard in collaboration with HHMI Janelia Research Campus, a biomedical research center in Virginia.reuters+2

"We built this for science to sort of show the promise of AI, but there's also huge benefits here for enterprise and for industry," Elizabeth Kelly, head of beneficial deployments at Anthropic, told CNBC.cnbc

The company says MHS works with any device that has a programmable interface and reduces integration time from weeks or months to hours or minutes. It is model-agnostic, meaning it functions with any large language model, not just Anthropic's Claude family. The standard builds on Anthropic's Model Context Protocol, an open standard for connecting AI agents to software data sources that the company released in 2024.fortune+2

Industry Partners Line Up

A range of organizations have signed on as early partners. Amazon Web Services Amazon.com, Inc. ↘ 0.35% will support MHS through its Strands Robots library, while Universal Robots, Doosan Robotics, Tecan, Danaher, QIAGEN, and Genentech are among companies testing or adding MHS support to their equipment. Carnegie Mellon University and quantum computing company QuEra also have early access.investing+1

Alek Kemeny, a member of Anthropic's technical staff, told Fortune that the company is working with device manufacturers to build new products pre-loaded with MHS and to retrofit existing hardware.fortune

Safety and Open-Source Plans

Anthropic said it is sharing the early version with partners to develop safety evaluations before making the standard open source. The framework includes guidelines for how AI agents should and should not interact with hardware, addressing concerns about equipment damage or misuse.wired+1

The announcement comes amid broader industry interest in combining AI with robotics. Nvidia ↗ 2.78% CEO Jensen Huang predicted in March that "every industrial company will become a robotics company," and Anthropic itself is building a silicon team and recently hired hardware executive Caitlin Kalinowski, who previously worked at OpenAI, Meta ↗ 0.84%, and Apple ↘ 0.52%.cnbc+1

Jonah Cool, head of partnerships and deployment of science at Anthropic, said much existing scientific equipment "suffers from proprietary solutions that are very brittle and often don't meet the need of scientists." He added: "We want to avoid vendor lock-in for scientists".fortune