Anthropic buys 1 person drug startup, eyes trials by 2028

Updated Oct 30, 2026, 11:58 AM CT: adds how the mouse study was run and the Claude errors Anthropic’s technical report examined.

Mithril Medicine took a possible treatment for a rare form of muscular dystrophy from an idea to promising results in mice in less than 18 months, with one employee. Anthropic, the company behind the Claude AI model, said on Oct. 27 that it bought Mithril and will run the program itself.

Mithril’s founder, Cody Tranbarger, used Claude to organize contracts, logistics and experiments with 11 outside labs, Anthropic said. He handed many other parts of running the company to Claude. He now works at Anthropic and plans to start several new programs. If his drug candidates keep holding up, Anthropic expects one to be ready for human trials as soon as late 2028.

Anthropic did not say what it paid. Mithril is working on two rare forms of muscular dystrophy that Anthropic describes as childhood diseases. It did not name them.

What six drug developers reported

Anthropic also described several months of work with six drug developers: BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines and The Column Group. The numbers below come from Anthropic’s post and the partners’ technical reports it links to. None of them comes from a human trial.

• Enveda. Claude reviewed 1,061 pages of drug manufacturing records. It caught all 11 problems human reviewers had found and flagged 7 they missed. Review time fell 89%, from over 25 hours to under 3. Claude drafted one regulatory document in 4 hours that takes about 3 months by hand.
• Divergence Labs. For a food allergy drug, Claude produced about 35,000 antibody designs in about a month. In early lab tests, 3 of 20 designs that pulled the allergy molecule IgE off immune cells did it better than the standard medication.
• Generate Biomedicines. Working on CAR T cell therapy for solid tumors, Generate scientists used Claude to design a piece of the therapy’s genetic code about 70% shorter than the standard one. In lab tests, immune cells made roughly 10 times as many cancer spotting receptors.
• BigHat Biosciences. Its lab made and tested about 1,010 versions of two antibody fragments in about 10 weeks. The best versions bound more than 10 times more tightly than the originals.

Bristol Myers Squibb is using Claude to weigh plans for an early stage drug already in clinical trials and to estimate the odds that those trials succeed. The two companies plan to expand that work in 2027.

What it means for patients

Nothing in this report is a new medicine you can get. Anthropic says the work is mostly preclinical, meaning the drugs have not been tested in people. A result in mice is not proof a drug works in a child.

The mouse study was small. Anthropic’s technical report is not peer reviewed. Seven mice given one version of the antibody had about twice the grip strength and peak force of 4 mice given a control solution. Each litter got one treatment, and the study was not randomized. Researchers tested the treated mice 126 and 161 days after the two control sessions, and the colony moved to a new facility in between. The report calls for a blinded repeat. The team’s largest screen covered 113 antibody constructs. It also examined 6 errors Claude made. At least 3 reached outside labs in emails people had edited and sent. Two went unnoticed for more than 4 days, R&D World reported on Oct. 27.

Anthropic’s case is about speed and cost. It says the slow parts of drug development are spread across the whole process: finding a target, designing and improving a molecule, manufacturing it, writing regulatory filings and running trials. The Enveda and Mithril numbers are aimed at the work after the first design.

The company also says investors favor safer, already tested ideas because drug discovery is slow and expensive. Its bet is that cheaper programs would let them fund riskier ones. Rare childhood diseases are on its list of targets.

The test is a human trial. By Anthropic’s own timeline, the earliest one from Mithril starts in late 2028.

What to watch

Anthropic says it will push Claude further into clinical trials as these programs advance and as it signs partners with drugs already in human testing. It is also giving Claude credits to life scientists and has opened a program that lets vetted research groups use its strongest models with looser safety limits for biology work.

Families dealing with a rare muscular dystrophy can search ClinicalTrials.gov by sponsor to see when a Mithril or Anthropic study opens.