On October 27, 2026, Anthropic published an update on using Claude in drug development. It names six partners: BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines and The Column Group. It also says Anthropic has "begun to run some research programs ourselves". A day later, on October 28, it expanded its AI for Science program to foundations, institutes and research projects.
A one-person biotech joins Anthropic
Anthropic acquired Mithril Medicine, "a one-person company that has been developing treatments for two rare forms of muscular dystrophy". Its founder used Claude "to organize contracts, logistics, and experiments with 11 external labs". Anthropic says Mithril took a possible treatment from an idea to promising results in mice in less than 18 months. The forecast is conditional: "If his drug candidates continue to hold up, we expect a drug to be ready to start clinical trials in humans as soon as late 2028."
The interesting part is the operating model. The lab work was done by contract labs. Claude coordinated the work between them. That makes the model a project manager across organizations as much as a scientific tool.
What the partners report
The results are specific but early. Divergence Labs had Claude generate about 35,000 antibody designs in roughly a month. BigHat made and tested about 1,010 variants in about ten weeks. Generate Biomedicines reports a CAR T-cell component design about 70% shorter than the standard version that performed better in lab tests. Enveda reports that on 1,061 pages of documentation Claude caught every issue human reviewers had found plus some they had missed, and that review time fell 89%. Bristol Myers Squibb is using Claude to assess clinical trial plans.
Credits flow through foundations
The October 28 expansion changes how the AI for Science program reaches researchers. The program, which started in May 2025, "gives free Claude credits to researchers working on high-impact scientific projects". Funders including the Alfred P. Sloan Foundation, the Simons Foundation, Schmidt Sciences and the US National Science Foundation will distribute credits to their own researchers. Institutions such as the Wellcome Sanger Institute, the Francis Crick Institute, EMBL and Gladstone Institutes are partners, as are projects including FlyWire and gnomAD. The post gives no credit amounts.
What remains uncertain. Anthropic says "all of this research is still at an early stage" and mostly preclinical. Results in mice often fail to carry over to humans, and the late-2028 date depends on candidates holding up. Most partner results are reported by Anthropic and the partners, not in peer-reviewed papers. Anthropic now both sells the model and runs drug programs with it, so its claims about Claude's value in drug discovery are no longer from a neutral supplier.
Analysis: Strip away the biology and this is a claim about long, multi-party work: one person plus a model coordinating eleven outside labs over more than a year. That is the same shape as running a supplier network, and the risks are the same. Every handoff needs a record of what was asked, what came back and who checked it. If Belgavi's Book Store let an agent coordinate its printers and couriers for a new edition, the record of decisions would matter more than the speed of any single step. For science, it also matters for reproducibility. A drug program run partly by a model should be auditable later, step by step.
Sources and evidence
On October 27 Anthropic reported early results from six drug-development partners and said it acquired Mithril Medicine, a one-person biotech whose founder used Claude to coordinate 11 external labs and reach promising mouse results in under 18 months (human trials possible as soon as late 2028 if candidates hold up). On October 28 it extended AI for Science credits through funders such as Sloan, Simons and NSF; Anthropic says the work is early and mostly preclinical.