--- title: "Anthropic unveils the MHS standard: AI agents gain direct access to physical hardware" description: "Anthropic has unveiled an experimental MHS standard that allows AI agents to directly control physical hardware and independently conduct experiments without complex software." date: 2026-08-28T13:34:06.000Z lang: en url: https://xab.info/en/posts/anthropic-mhs-standard-ai-agents-physical-hardware tags: [anthropic, artificial-intelligence, mhs, ai-agents, robotics, open-source] publisher: "XAB.info" --- # Anthropic unveils the MHS standard: AI agents gain direct access to physical hardware ![Typewriter with ARTIFICIAL INTELLIGENCE typed on paper symbolizing Anthropic's MHS standard giving AI agents direct access to physical equipment](https://xab.info/media/2026/08/28/anthropic-mhs-standart-dlya-ii-agentov/anthropic-mhs-standart-dlya-ii-agentov-1.webp) ## 🎯 Key Points - MHS creates a unified interface and data exchange format for heterogeneous physical devices without specific translator programs. - Combined with the Model Context Protocol, scientists control equipment in natural language, while the AI adjusts parameters in real time. - In the demonstrations, Claude independently calibrated a laser, worked with a microscope, and controlled a robotic arm without step-by-step algorithms. - The standard is currently in a research preview; partners include AWS, Hugging Face, Raspberry Pi, Automata, and Universal Robots, with open-source plans on the horizon. Anthropic has unveiled an experimental standard called MHS, which gives AI agents direct access to physical hardware. According to the company, the standard creates a unified interface and data exchange format that allows heterogeneous devices to communicate over a network without the need for individual translator programs. This moves autonomous models from a virtual environment into a real laboratory, where they can independently conduct experiments, operate instruments, and adjust parameters in real time. ### What is MHS and how it works The standard is built on the idea of unification: instead of developing a specific driver for each device, MHS defines a common data exchange protocol. A device describes its capabilities and limitations in a standardized format, after which it becomes "readable" to a language model. This approach removes one of the main barriers to automation — the need to write complex software for every new instrument. ### Language instead of code: integration with Model Context Protocol Combining MHS with the Model Context Protocol allows scientists to issue commands to equipment in plain natural language. The AI model, in turn, analyzes the current situation, adjusts parameters in real time, and resolves hardware malfunctions. In effect, the human acts as the operator-client, while routine control and monitoring are handed over to the model. ### Demonstrations: calibration, microscope, and robotic arm During the demonstrations, a Claude-level model independently calibrated a laser, verifying the result through a separate camera and fine-tuning the optical system. In another scenario, the AI focused a microscope, analyzed the image, identified an area of interest, and independently shifted the objective lens to continue the analysis. Particularly notable was the episode with the robotic arm: the model figured out how to lift an aluminum can without step-by-step algorithms from a human. ### Standardized tags and safety Because the models were trained primarily in a virtual environment, MHS includes a system of standardized tags describing the physical limitations of the "hardware." This allows the AI to instantly read the capabilities of an unfamiliar device. At the research preview stage, the company is focusing on safety assessment and the development of best practices for AI working with real equipment. ### Partners and open-source plans In the research preview, Anthropic is collaborating with a number of laboratories and manufacturers, including Amazon Web Services (Strands Robots), Hugging Face (LeRobot), Raspberry Pi, Automata, and Universal Robots. In the future, the company plans to make MHS a fully open standard (open-source), independent of specific AI model developers, which could accelerate the automation of scientific activity as a whole. ## 🔍 Fact-Check Verification - [AI will no longer just sit in a chat: Anthropic gave it access to real hardware](https://www.rbc.ua/ukr/news/shi-bilshe-siditime-chati-anthropic-dala-1787912405.html) - Подтверждает анонс стандарта MHS и доступ ИИ к физическому оборудованию. - [Anthropic gave Claude a physical body (almost) — meet MHS](https://www.securitylab.ru/news/576597.php) - Описывает MHS как связку Claude с реальным «железом», согласуется с демонстрациями. - [Microsoft, OpenAI, and Anthropic accelerate the race for "AI agents" that could replace corporate ...](https://www.ixbt.com/news/2026/06/18/microsoft-openai-anthropic.html) - Контекст: общая гонка разработчиков за автономными ИИ-агентами, усиливает значимость анонса. - [Research showed that AI agents have already learned to win people's trust better than ...](https://3dnews.ru/1146022/issledovanie-pokazalo-chto-iiagenti-uge-nauchilis-luchshe-vtiratsya-v-doverie-k-lyudyam-chem-realnie-moshenniki) - Подтверждено по источнику 3dnews.ru ## ❓ FAQ ### Q: What is the MHS standard from Anthropic? **A:** It is an experimental standard that creates a unified interface and data exchange format, allowing heterogeneous physical devices to communicate over a network without the need for individual translator programs. ### Q: How do scientists control equipment through MHS? **A:** Combined with the Model Context Protocol, commands are issued in natural language, while the AI model analyzes the situation, adjusts parameters in real time, and resolves hardware malfunctions. ### Q: Will MHS be open? **A:** According to Anthropic's plans, MHS will eventually become a fully open standard (open-source), independent of specific AI model developers; it is currently in the research preview stage.