Anthropic, the artificial intelligence company behind Claude, has quietly moved beyond software and into a physical biology laboratory, marking a significant expansion of its ambitions in drug research.
The company has established a “wet lab” in the San Francisco Bay Area, where scientists can run real biological experiments rather than relying only on computer simulations. Anthropic’s head of life sciences, Eric Kauderer-Abrams, confirmed the facility to Reuters, saying physical laboratory work remains essential for testing biological ideas.
The lab is not being described by Anthropic as a dedicated drug-discovery facility. However, it sits inside a broader push by the company to use Claude and other AI tools to accelerate research into medicines, particularly for rare and difficult-to-treat diseases.
Proposal for Conducting Cyber Crisis Drill, Tabletop Exercise (TTEx) & CCMP Readiness Exercise
From chatbot to laboratory assistant
Anthropic wants Claude to do more than analyse papers or write research code. Reuters reported that the company is exploring how its AI models could direct robotic laboratory equipment to carry out experiments with limited human intervention.
That could mean an AI system helping decide what experiment should happen next, then sending instructions to machines that handle samples, liquids or testing equipment. Anthropic says human oversight will remain important, particularly because biological research can involve serious safety risks.
The move follows the June launch of Claude Science, an AI workbench built for researchers. Anthropic says the platform can connect scientists with databases, computing tools and research software that would normally require switching between multiple systems.
Anthropic has also been building infrastructure around this idea. In August, it launched what it calls the Model Hardware Standard, designed to help AI systems interact with physical laboratory equipment. The company has meanwhile expanded access to advanced models for verified life-sciences organisations.
What “wet lab” and “in silico” actually mean
A wet lab is simply a laboratory where researchers work with real biological or chemical material. Scientists may test cells, proteins, compounds or samples using physical equipment.
“In silico” means the opposite side of that workflow: experiments or predictions performed on computers. AI can rapidly suggest promising molecules or biological targets, but those predictions still need to survive testing in the real world.
That distinction matters because drug development is full of failures. A molecule may look promising on a computer and still prove ineffective or unsafe once it is tested in cells, animals or humans.
Anthropic’s strategy appears to be to connect both sides. Claude can help generate or analyse ideas digitally, while laboratory work can test whether those ideas actually hold up.
AI companies move deeper into drug development
Anthropic is not entering an empty field. Companies such as Recursion have spent years combining automated wet labs with machine learning to generate biological data and identify potential drug candidates.
Recursion says its platform can run large-scale experiments while AI models analyse the resulting biological and chemical data. Its programmes have also moved into clinical development, showing how far the AI-drug-discovery sector has progressed beyond simple software demonstrations.
Anthropic is taking a more limited approach for now. Kauderer-Abrams told Reuters that the company does not plan to run clinical trials itself and wants to avoid competing directly with pharmaceutical companies.
Instead, it is positioning Claude as a research engine and partner. Earlier this week, Novo Nordisk announced that it would test Anthropic’s models and Claude Science in drug research and development. Novo already uses Claude for some documentation tasks and says AI could also support deeper scientific reasoning.
Anthropic has said it wants to focus partly on diseases that traditional pharmaceutical companies may consider commercially unattractive or technically difficult. That includes so-called “undruggable” conditions, where conventional approaches have struggled to identify effective treatments.
The opportunity is large, but so are the risks. Anthropic has itself warned about advanced AI systems potentially being misused in biology. The company says its life-sciences work therefore requires a balance between making models more capable and controlling how those capabilities are deployed.
For patients and researchers, the real test will not be whether AI can suggest more molecules or automate more laboratory steps. It will be whether those systems can help produce treatments that prove safe and effective in years of clinical testing.
What this means for you: AI may speed up early-stage drug research, but an AI-designed treatment still has to pass the same safety and clinical testing required of other medicines. Treat claims of “AI-discovered cures” cautiously until human trial evidence exists.
Follow for daily updates on cybercrime, corporate fraud, DFIR, hacking, investigations, and digital forensics
