Anthropic has launched a new wet lab facility in the San Francisco Bay Area dedicated to biological research. The company plans to utilize robots to automate scientific tasks, with operations controlled by its Claude series of large language models.
The facility will implement the Model Hardware Standard (MHS) protocol, which allows AI models to manage laboratory equipment, detect experimental errors, and generate automation scripts. This protocol was developed in collaboration with the Virginia-based HHMI medical research center.
During testing of the Claude Mythos 5.1 model, the system generated 12 candidate molecules for high-affinity binders. These molecules achieved a 50% hit rate for therapeutic potential, significantly higher than the typical 10% to 15% rate seen in standard protein design projects.
While the company has not disclosed specific research projects, it confirmed the lab is not focused solely on drug discovery. Anthropic intends to work with external partners and may license its findings to pharmaceutical companies rather than bringing products to market independently.
