Anthropic's first drug results with Claude: six partners, one 25-hour review cut to 3 hours
Anthropic says Claude helped redesign a cancer therapy and drew up 35,000 antibody designs in a month. Almost none is tested in people yet.
On 27 October 2026 Anthropic published early results from research partnerships with six drug developers, plus programmes of its own. Its argument is that drug development is slow at many steps, not just at designing the molecule, so it is trying Claude at each of them.
Designing molecules. Six scientists at Generate Biomedicines used Claude to redesign part of a CAR T-cell therapy, a cancer treatment made from a patient’s own immune cells. Their version was about 70% shorter than the standard part, and in lab tests the cells grew roughly 10 times as many of the receptors they use to spot cancer. For Divergence Labs, working on food allergies, Claude drew up about 35,000 antibody designs in a month; early tests found 3 that stuck to the target molecule. BigHat Biosciences made and tested about 1,010 versions of two drug candidates in about 10 weeks, and the best stuck to their target more than 10 times more tightly.
Paperwork. At Enveda, Claude read 1,061 pages of manufacturing records. It found all 11 problems the human reviewers had found, plus 7 they had missed, and the review took under 3 hours instead of more than 25. Bristol Myers Squibb uses Claude to assess clinical trial plans.
Its own work. Anthropic bought Mithril Medicine, a one-person company working on two rare forms of muscular dystrophy. Its founder used Claude to coordinate experiments at 11 outside labs and got one treatment from idea to promising mouse results in under 18 months. Anthropic says human trials could start as soon as late 2028.
What it means
The most solid number here is the dullest one: a records review that took over 25 hours took under 3, and caught more problems. Paperwork and coordination are a real part of why medicines are slow, and that is where the evidence is clearest. The molecule results are early lab work from companies that gain if Claude looks good, and no patient has been treated, so nothing here shortens the wait for a new drug yet.
Caveats
- These are reports from Anthropic and its partners, who all gain if Claude looks useful. We found no independent replication.
- Anthropic says the work is mostly preclinical: it has not been tested in people. Most drug candidates that look good in the lab or in mice fail in human trials.
- A “hit” in an early lab test is a first step. Divergence Labs has not yet built the two-part drug its plan needs.
- Only Enveda’s review times come with a before-and-after comparison. The other results do not show what the same teams would have achieved without Claude.