Anthropic Sets Up Biology Lab for AI Drug Research

Artificial intelligence company Anthropic has established a physical biology laboratory in the San Francisco Bay Area, marking a new step in its effort to apply AI to scientific research and drug development.

The company has increasingly moved its life sciences work beyond computer-based analysis. Reuters reported on September 18 that Anthropic has built a “wet lab” where physical biological experiments can be conducted, citing people familiar with the matter and comments from Eric Kauderer-Abrams, Anthropic’s head of life sciences.

Anthropic has said its life sciences programme is aimed at using AI to accelerate scientific discovery, including research into diseases that may receive less attention from traditional pharmaceutical companies because of their commercial economics.

The company has not publicly disclosed a specific drug pipeline from the new laboratory, and Anthropic said the facility should not be described as being specifically for drug discovery. Reuters reported that the lab will instead support broader physical biology work as the company’s life sciences programme expands.

Why Anthropic is moving into physical biology

AI models can analyse scientific literature, biological datasets and molecular structures at high speed, but biological hypotheses ultimately have to be tested in the real world.

Anthropic’s Kauderer-Abrams told Reuters that real laboratory work remains the final test for biological research and confirmed that the company is conducting such work itself as well as through external partners.

The move represents a shift from relying primarily on computational research toward combining AI systems with physical experimentation.

That distinction matters because a promising computer-generated prediction is not the same as a validated biological result. Experimental work is needed to determine whether a molecule, protein or other biological design behaves as expected.

Anthropic has already been developing tools intended to support scientists through several stages of this process.

Claude Science is becoming part of Anthropic’s research strategy

In June, Anthropic launched Claude Science, an AI workbench designed for scientific researchers.

The company says the system integrates commonly used scientific tools and databases, produces auditable research outputs and provides access to computing resources. Its stated goal is to reduce the fragmentation researchers face when working across databases, programming environments and specialised scientific software.

Anthropic has also expanded Claude’s connections to life sciences platforms.

Its life sciences tools include connections to resources such as PubMed, Benchling, ClinicalTrials.gov, ChEMBL and Open Targets, covering areas including biomedical literature, clinical trials, bioinformatics and potential drug targets.

This gives Anthropic a broader presence in scientific workflows than simply providing a general-purpose chatbot.

Anthropic has already demonstrated biology capabilities

The company has been publishing examples of Claude’s performance on scientific tasks.

In August, Anthropic said Claude had designed protein binders against 15 targets in an experimental study, with successful binding reported for 14 of the targets. The company said between 22% and 35% of individual designs bound successfully depending on the experimental setup, compared with what it described as a typical 10% to 15% rate in protein-design campaigns.

Anthropic also reported that Claude completed certain analytical chemistry tasks substantially faster than the contract laboratory’s own analysis while producing results that closely matched the laboratory’s measurements.

These are company-reported research results, however, and they do not demonstrate that AI has independently developed an approved medicine.

Drug development remains a long process involving laboratory validation, toxicity assessment, clinical trials, regulatory review and manufacturing.

New partnerships are expanding Anthropic’s reach

Anthropic’s move into physical biology comes as it builds relationships with pharmaceutical and research organisations.

On September 16, Novo Nordisk announced a collaboration with Anthropic to use Claude and Claude Science for scientific research and drug discovery.

The Danish pharmaceutical company said the partnership would focus on specific scientific challenges identified by its researchers and computational teams and support biological reasoning and the development of new medicines.

Anthropic has also worked with organisations including the Allen Institute and Howard Hughes Medical Institute on scientific research.

The company says these partnerships are intended to help researchers handle increasingly large volumes of biological data and accelerate the process of turning information into validated scientific insights.

A $400 million acquisition adds to the strategy

Anthropic has also strengthened its life sciences capabilities through acquisitions.

Reuters reported that the company acquired Coefficient Bio for about $400 million in stock, adding technology and expertise to its drug-development efforts.

The acquisition comes alongside the development of Claude Science and Anthropic’s growing network of pharmaceutical and biotechnology relationships.

Together, these moves suggest the company is building a broader scientific infrastructure around its AI models rather than treating biology as simply another chatbot use case.

Anthropic is also tightening access to advanced biology capabilities

The expansion into biological research comes with significant safety concerns.

On September 17, Anthropic announced its Life Sciences Verification Program, giving vetted life sciences organisations access to models with safeguards designed specifically for advanced biology-related work.

The beta programme initially targets teams and institutions and is intended to support areas including drug discovery, research biology, clinical development and manufacturing.

Anthropic has said some advanced biological capabilities can have both beneficial and harmful applications.

The company has therefore placed stronger restrictions on certain biology-related uses of its generally available models while creating trusted-access programmes for professional researchers.

That approach reflects a central problem in AI-enabled biology: the same capabilities that can help researchers design medicines may also have potential misuse risks.

Human oversight remains part of the laboratory approach

The use of AI and robotics in physical laboratories could eventually allow researchers to automate portions of experimental work.

Reuters reported that Anthropic is exploring whether Claude can instruct robotic systems to carry out laboratory experiments with limited staffing.

The company has stressed that humans remain involved because of the safety requirements surrounding biological experimentation.

This does not mean Anthropic is operating a fully autonomous laboratory.

Instead, the emerging model involves AI assisting with research tasks while humans remain responsible for experimental decisions, supervision and interpretation.

The company is targeting areas that may be overlooked

Anthropic has said it sees particular potential in diseases that pharmaceutical companies may overlook because the commercial returns are uncertain.

That strategy could allow AI-assisted research to be applied to areas where better computational tools reduce some of the cost and time involved in early-stage research.

But identifying a promising biological target is only one part of drug development.

A candidate treatment must still demonstrate safety and effectiveness through increasingly demanding stages of testing before it can reach patients.

Anthropic’s current work should therefore be viewed as an effort to accelerate parts of the research pipeline rather than evidence that AI has removed the traditional barriers to developing medicines.

The wider AI race is moving into science

Anthropic’s biology laboratory is part of a broader technology-industry push to apply advanced AI systems to scientific research.

Companies are increasingly using AI to analyse biological data, predict molecular structures, identify possible drug targets and automate laboratory tasks.

Anthropic’s own research programme has expanded quickly in 2026, including its Claude Science platform, partnerships with major research organisations, specialised models for biology and the new Life Sciences Verification Program.

The new physical laboratory adds another component to that strategy.

For Anthropic, the challenge will be to show that its models can produce useful experimental results consistently enough to justify the cost and complexity of combining advanced AI with laboratory science.

For the wider pharmaceutical industry, the more important question may be whether these systems can shorten research timelines while producing results that withstand independent scientific and regulatory scrutiny.

Anthropic’s new laboratory does not answer that question yet.

It does, however, show that the company is moving from using AI to analyse biology toward using AI alongside physical experiments, bringing its ambitions further into the real-world process of scientific discovery.