Anthropic published a wide-ranging update on Tuesday detailing its push into drug development, and the standout number is 35,000. In a partnership with Divergence Labs aimed at food allergies, the company says Claude converted research goals into a detailed design plan and then, largely autonomously, generated roughly 35,000 antibody designs in about a month for the lab to narrow down and test.
The update, posted on Anthropic's research blog, describes deep partnerships with six organizations developing new medicines — BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines, and The Column Group — along with the company's first in-house drug discovery efforts, including a small acquisition. For more context on this story, see our ongoing breaking AI news.
Hits in the Food Allergy Program
Divergence Labs is pursuing a different approach to treating food allergies. Allergic reactions are set off by a molecule called IgE, some of which sits attached to immune cells and some of which floats freely in the blood. Standard medications neutralize the circulating IgE; Divergence Labs wants drugs that detach IgE from immune cells entirely, which could be far more effective but demands a precise two-part design: one arm of the molecule must bind tightly to IgE while the other pries it away from the immune cell.
According to Anthropic, Claude turned that goal into a design plan and produced the roughly 35,000 antibody candidates in about a month. In early laboratory tests, the program found hits for both arms: three designs that could bind to IgE and 20 that could dislodge it. Three of those 20 appeared to dislodge IgE more effectively than the standard medication, and Divergence Labs' next step is building a drug that combines both arms.
Six Partnerships Across the Drug Pipeline
The partnerships map onto different stages of drug development. At Generate Biomedicines, work centers on reengineering CAR T-cell therapy — historically successful against blood cancers — so it can treat solid tumors, in part by shortening one segment of the therapy's genetic instructions to make room for new ones. BigHat Biosciences, which pairs machine learning with an automated lab, worked with Claude to improve binding, stability, and manufacturability in two antibody-fragment leads.
The less glamorous stages matter too. At Enveda, manufacturing chemists were spending up to 75 percent of their time on paperwork; the company built a system connecting all of its manufacturing documents to Claude, which flags potential issues in batch records needed for clinical trials and regulatory submissions. At The Column Group, the collaboration explores whether cutting the time and cost of discovery can make investors more willing to fund ambitious early-stage ideas. Bristol Myers Squibb, the most recent partner, is using Claude to assess clinical trial plans and estimate their likelihood of success, with an expansion planned for 2027.
The initiative is led by Anthropic's life sciences team, whose stated strategy is to embed AI across every stage rather than chase any single bottleneck. "There is no single problem to solve; the bottlenecks are distributed all throughout the drug development process," the company wrote in the post, arguing that "the best way to identify and work through these bottlenecks is to get deeply involved in real drug programs, across many different disease areas and stages of development." The team says it has already prioritized specific model improvements based on what the partnerships have revealed about Claude's strengths and limitations in laboratory settings.
Anthropic Bought a One-Person Drug Company
The most unusual disclosure is that Anthropic acquired Mithril Medicine, a one-person company founded by Cody Tranbarger that was developing treatments for two rare forms of muscular dystrophy. Tranbarger used Claude to organize contracts, logistics, and experiments across 11 external labs, delegating many operational aspects of the company to the model.
The result, per Anthropic: Mithril moved a possible treatment for one form of muscular dystrophy from an idea to promising results in mice in less than 18 months. Tranbarger has joined Anthropic, and if the drug candidates continue to hold up, the company expects a candidate ready to begin human clinical trials as soon as late 2028.
Still Early, Still Preclinical
Anthropic is explicit that the work remains at an early stage and is mostly preclinical — for the most part, these drug programs are not yet being tested in humans. The company says the partnerships have already driven specific improvements to its models and that it plans to extend Claude's role into clinical trials as programs mature.
The update arrives as AI labs compete to prove their models can do real scientific work rather than just answer questions. Anthropic has rolled out Claude Science, an AI workbench for researchers, alongside life sciences credits and a Life Sciences Verification Program that gives vetted organizations access to its most powerful models with more permissive biology safeguards.
If the early results hold, the significance is less about any single drug candidate and more about the shape of the workflow: a small biotech generating tens of thousands of candidate designs in a month, with an AI system handling much of the scientific and operational load, compresses timelines that traditionally run in years. Whether those designs survive contact with wet-lab reality and, eventually, clinical trials will be the real test — but for now, Anthropic has put concrete, testable numbers behind the claim that AI can accelerate the earliest, most speculative phase of drug development.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
