Anthropic has established a biology laboratory, aiming to enable the Claude AI system to control physical robots.

Beating AI News Flash: Anthropic has already built its own biological laboratory in the San Francisco Bay Area, taking Claude from biological research on computers to real experiments. Eric Kauderer-Abrams, Anthropic's head of life sciences, confirmed that the company now both conducts experiments itself and delegates part of the work to external partners.

The next step is to have Claude directly participate in laboratory automation. Anthropic hopes Claude can command robotic arms, pipetting robots, and other equipment to carry out experiments on its own under human supervision. Last month, the company launched the Model Hardware Standard, allowing AI to control microscopes, robotic arms, and laboratory instruments through a unified interface. It can already adjust parameters based on experimental results and even handle some equipment failures.

Anthropic has also begun conducting its own preclinical research related to drugs, focusing on diseases that traditional pharmaceutical companies consider too difficult or not commercially valuable enough. However, the company will not conduct human clinical trials for now, and it also said its laboratory is not specifically intended for drug discovery, hoping to avoid directly competing with pharmaceutical company clients.

Life sciences is already one of the areas where Anthropic has invested the most manpower and resources. In June of this year, the company also launched Claude Science, putting scientific research tools such as paper databases, code, and computing resources into one workbench. Now, with the addition of a real laboratory, Claude's scientific research is continuing to advance from "analyzing data" to "conducting experiments hands-on."