Anthropic published "Early progress on accelerating drug development with Claude" on October 27, 2026. The post covers deep research partnerships with six organizations (BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines, and The Column Group) plus Anthropic's own drug-discovery efforts after acquiring Mithril Medicine. The company says the work is still early and mostly preclinical.
Key Facts
| Details | |
|---|---|
| Published | October 27, 2026 (Anthropic research blog) |
| Partners named | BigHat Biosciences; Bristol Myers Squibb; Divergence Labs; Enveda; Generate Biomedicines; The Column Group |
| Internal programs | Rare childhood muscular dystrophy work via acquired Mithril Medicine (founder Cody Tranbarger); Anthropic says idea-to-mouse results in under 18 months with Claude coordinating much of the scientific and operational work across 11 external labs |
| Stated aim if candidates hold | Human clinical trials as soon as late 2028 (Anthropic's wording; conditional) |
| Stage | Mostly preclinical; not yet testing most programs in humans |
| Related tools | Claude Science workbench; Claude credits for life scientists; Life Sciences Verification Program; open call for guest investigators |
Background
Anthropic's thesis is that bottlenecks sit across the pipeline (target choice, molecule design, manufacturing scale-up, regulatory filings, trial design), so Claude should be embedded in real programs rather than only molecule suggestion. Partner case studies in the post include:
- Generate Biomedicines: Claude Code used to propose more compact CAR T genetic designs; one design about 70% shorter than the standard version showed higher receptor expression and stronger lab tumor-cell killing in Generate's tests (company technical report; early lab results).
- Divergence Labs: food-allergy antibody concept with two "arms"; Claude generated roughly 35,000 antibody designs in about a month; early hits for binding and dislodging IgE reported.
- BigHat Biosciences: Claude connected to BigHat's Reccy lab system proposed variants; about 1,010 variants made/tested in about 10 weeks; best versions claimed >10x tighter binding and large stability/clumping improvements versus starting leads.
- Enveda: Claude-assisted manufacturing document review on 1,061 pages caught all 11 human-found issues plus seven more, with reported 89% less human review time; also first drafts of regulatory documents.
- The Column Group: establishing Pleiades, a new company intended to run more/complex programs with Claude; first target-assessment skill evaluated on 45 target-disease pairs.
- Bristol Myers Squibb: Claude for early clinical-trial plan assessment and success-likelihood estimates; expansion planned in 2027.
On Mithril, Anthropic's own post states the July acquisition path and the late-2028 human-trial readiness expectation if candidates continue to hold. R&D World and Unite.AI, summarizing Anthropic's materials, note the accompanying LAMA2-CMD mouse technical report is not peer reviewed and that the pilot mouse study had design limits (e.g. non-randomized litter assignment and staggered testing in R&D World's reading of the report). Those caveats belong with the mouse numbers; Anthropic itself frames the results as early proof-of-principle, not patient efficacy.
What We Don't Know Yet
- Which partner programs, if any, will reach IND-enabling studies or first-in-human dosing on what calendar
- Peer-reviewed publication outcomes for the partner technical reports and the Mithril mouse study
- Deal terms of the Mithril acquisition (not disclosed in the Anthropic post)
- How far Claude Science credits and the Verification Program will change industry practice outside these case studies