Anthropic acquires one-person pharma company Mithril Medicine and partners with six institutions to advance AI drug discovery

Anthropic is extending AI capabilities from conversational tools into the core processes of drug discovery.

According to the latest update published on Anthropic's official website on October 27, the company's life sciences team has established in-depth research collaborations with six institutions and has acquired Mithril Medicine, a biotechnology company run by a single person, to directly enter drug discovery.

Anthropic said that its AI model Claude has demonstrated quantifiable efficiency gains in multiple areas, including molecular design, manufacturing document review, regulatory filings, and clinical trial planning, and some early data has been disclosed publicly.

The company explicitly stated that it plans to "substantially expand" the scale of this work and will continue to extend Claude's involvement into the human clinical trial stage.

Acquiring a "one-person pharma company" and doing drug development firsthand

Anthropic's most notable move this time is the acquisition of Mithril Medicine—a tiny biotechnology company independently operated by founder Cody Tranbarger.

Mithril Medicine focuses on research into treatments for two rare childhood muscular dystrophies. Tranbarger previously used Claude to coordinate contracts, logistics, and experimental arrangements with 11 external laboratories, and delegated a large amount of the company's operational affairs to Claude.

The result: in less than 18 months, Mithril Medicine advanced one of its candidate therapies for a muscular dystrophy from concept to the mouse experiment stage, achieving preliminary positive results.

Tranbarger has now joined Anthropic to continue advancing the research and plans to launch multiple new projects. Anthropic said that if the candidate drug continues to perform well, it is expected to be ready to initiate human clinical trials as early as the end of 2028.

Six major partners covering the entire R&D chain

Beyond its own research, Anthropic has over the past several months launched collaborations with six institutions, covering multiple stages of drug discovery from early discovery to clinical planning.

Generate Biomedicines: CAR-T therapy engineering

Generate Biomedicines is attempting to extend CAR-T cell therapy from blood cancers to solid tumors. Six scientists used Claude Code to propose new designs for key genetic instruction fragments of the therapy. One design was about 70% shorter than the standard version, and laboratory tests showed that immune cells produced about 10 times as many receptors, had higher tumor cell killing efficiency, and remained active for longer.

Divergence Labs: new therapy for food allergies

Divergence Labs is developing a novel drug for food allergies, with the goal of completely stripping IgE molecules off immune cells rather than merely neutralizing free IgE in the blood. Claude translated its R&D goals into detailed designs and, in about one month, generated nearly 35,000 antibody designs almost autonomously. In early screening, researchers found effective candidates targeting two functional "arms": 3 that can bind IgE, and 20 that can strip IgE off immune cells, of which 3 outperformed existing standard drugs.

BigHat Biosciences: molecule optimization

BigHat Biosciences combines AI with automated laboratories to optimize candidate drug molecules. In about 10 weeks, the team tested about 1,010 molecular variants. Midway through, Claude and the researchers re-analyzed the original data and adjusted the evaluation metrics and experimental protocols. Final results showed that the best version had more than 10 times improved binding affinity, a candidate molecule targeting cancer cells had 18°C higher thermal stability, and another candidate molecule aggregated more than 100 times more slowly.

Enveda: document review and regulatory filings

Enveda's manufacturing chemists previously spent as much as 75% of their working time on document processing. Enveda built a system that feeds manufacturing documents into Claude to automatically flag potential issues. In a test on 1,061 pages of documents, Claude identified all 11 issues found by human reviewers and additionally flagged 7 issues that humans had missed. Overall manual review time was compressed from more than 25 hours to less than 3 hours, a reduction of 89%.

In addition, Claude drafted two important regulatory documents. One of them took 4 hours, whereas completing similar work manually typically takes about 3 months.

The Column Group: building an "AI-native" pharma company

The Column Group (TCG) is working with Anthropic to explore whether reducing the time and cost of drug discovery can encourage investors to support riskier innovative projects. TCG is setting up a new company, Pleiades, which plans to run more and more complex drug projects with a small team. Pleiades has already developed its first Claude skill module for evaluating the research value of drug targets, and has completed evaluations on 45 target-disease pairs.

Bristol Myers Squibb: clinical trial planning

BMS is the newest of the six partners. BMS is using Claude to analyze trial protocols for early-stage clinical drugs and to assess the probability of trial success.

Michael Pourdehnad, Senior Vice President of Early Clinical Development and Head of Hematology and Cell Therapy at BMS, said: "We use Claude in the clinical trial stage to analyze proprietary data, develop clinical research plans, and assess the likelihood of study success. With just one prompt, Claude generated a complete preliminary analysis to help us evaluate different development options—work that would typically take days if done manually."

The two sides plan to expand their collaboration in 2027 to further evaluate AI's potential to improve early development success rates.

Strategic layout: open platform, broader reach

Anthropic said the above collaborative research is only part of its life sciences strategy, with the goal of opening Claude's capabilities to a broader range of drug R&D organizations.

To this end, Anthropic has launched several supporting initiatives:

Claude Science: an AI workbench for researchers;

Life Sciences Verification Program: allowing trusted institutions to use the most powerful models under more relaxed safety restrictions to support legitimate biological research;

Open call for visiting researchers: soliciting proposals from research teams that want to use Claude for drug discovery and other key research questions;

At the same time, providing Claude usage credits to life sciences researchers.

Anthropic emphasized that all current research is still at an early stage, and most has not yet entered human clinical trials. The company plans to continue extending Claude's involvement into the clinical stage as drug projects advance. Risk disclosure and disclaimer

Markets carry risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investment based on this is at the user's own risk.