Anthropic's Claude AI is now helping design new medicines

Anthropic, the company behind the Claude AI assistant, says Claude is now helping real drug programs. Over the past several months it has worked with six organizations developing new medicines. In one early lab test, a cancer-therapy design Claude proposed was about 70% shorter than the standard version and performed better. These are lab findings, not treatments anyone can get.

What Anthropic is doing

The six partners are BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines and The Column Group. Anthropic also started its own drug discovery efforts, including programs for rare childhood forms of muscular dystrophy, where Claude handles much of the scientific and operational work. The company says drug development has many slow points, so it wants AI involved in many tasks, not only designing molecules.

A shorter cancer therapy design

Generate Biomedicines wants to widen CAR T-cell therapy, which has historically helped treat blood cell cancers, so it can also treat solid tumors. Six Generate scientists used Claude Code to propose designs for one part of the therapy's genetic instructions. One design was about 70% shorter. In lab tests, immune cells made roughly 10 times as many receptors, the proteins they use to spot cancer cells. The design also killed tumor cells more efficiently, and the cells stayed active longer.

Food allergy designs at scale

Divergence Labs is working on a different way to treat food allergies. The goal is to pull a molecule called IgE off immune cells, instead of only neutralizing the IgE floating in the blood. That takes a two-armed design: one arm grips IgE and the other helps pry it loose.

Claude turned the lab's goals into a design plan, then largely on its own made roughly 35,000 antibody designs in about a month. Early tests found three designs that could bind to IgE and 20 that could dislodge it. Three of those 20 appeared to dislodge IgE more than the standard medication. The next step is building a drug that combines both arms.

Testing in an automated lab

BigHat Biosciences pairs AI with an automated lab to improve potential drugs. Working with Claude, it made and tested about 1,010 variants of two drug candidates in about 10 weeks. Partway through, the team suspected the standard measures of heat stability and clumping were misleading. Claude and an Anthropic researcher reanalyzed the raw data and proposed different measures. The material available here cuts off before the final results.

What is not known yet

Anthropic says it is optimistic that some projects could become effective medicines for food allergies, rare childhood diseases and several cancers, but only if their positive trajectories continue. No timeline is given. The company also plans to scale this work up substantially. Details on the work with Bristol Myers Squibb, Enveda and The Column Group were not included in the material available here.