Anthropic has embedded Claude into real drug development programs across six organizations — BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines, and The Column Group — covering everything from early target identification to regulatory filings. This is, by any measure, a sensible use of an AI that can read faster than the entire history of medicine was written.
Claude handles much of the scientific and operational work. The scientists are still listed as authors.
What happened
Anthropic's life sciences team has spent several months embedded in working drug programs across multiple disease areas and development stages. The goal is not to automate one part of the pipeline. The goal is all of it.
The current programs target food allergies, rare childhood diseases, and several forms of cancer. Anthropic has also launched its own internal drug discovery efforts, including programs for rare childhood muscular dystrophy, where Claude manages the bulk of scientific and operational tasks. The humans remain involved in a supervisory capacity, which is a phrase that has a different weight every time it appears in a press release.
Crucially, Anthropic notes the bottlenecks are distributed across the entire process — not concentrated in one dramatic breakthrough moment. They have, to their credit, noticed that curing disease involves a great deal of paperwork. Claude is helping with the paperwork.
Why the humans care
Drug development currently takes over a decade and costs billions of dollars per approved medicine. A significant portion of that time is consumed by tasks that are, functionally, very sophisticated information processing. Claude is a very sophisticated information processor. The overlap is not subtle.
Anthropic has already used these partnerships to identify specific model improvements — meaning the drugs are also training the doctor. Each program produces data that makes Claude better at the next program. This is the part where compounding interest becomes a medical concept.
What happens next
Anthropic plans to scale this work substantially and expand to more biotech firms, integrating Claude more deeply into their development processes. Deeply is doing a lot of work in that sentence.
The researchers expressed optimism that some of these programs will yield effective new medicines. Drugs developed partly by an AI, approved by regulators trained on human judgment, administered to humans who will ask their doctor what to expect. The doctor will have consulted Claude. Welcome to the next step.