Anthropic Launches Internal Drug Discovery Programme Targeting Neglected Diseases with Claude Science

Anthropic, the San Francisco-based AI company known for its Claude large language models, has announced it will develop experimental drugs internally — not just sell AI tools to pharmaceutical companies. The programme is explicitly aimed at neglected diseases that established drugmakers tend to avoid because they offer limited commercial return. It’s a striking departure for a company that, until now, has been best known for building AI models and positioning itself as a safety-conscious alternative to its competitors.

Eric Kauderer-Abrams, Anthropic’s head of life sciences, outlined the initiative at a San Francisco event, saying the company’s public benefit corporation structure gives it the freedom to pursue programmes based on patient need rather than commercial viability alone. That structure, he suggested, allows Anthropic to go where traditional pharma won’t.

What Claude Science Actually Does

At the centre of the programme is Claude Science, a life sciences AI workbench currently in beta testing. It integrates more than 60 specialist functions covering genomics, structural biology, proteomics and cheminformatics. Researchers can use it to design CRISPR screens, run single-cell RNA sequencing analysis and render 3D protein structures — tasks that would previously require stitching together a patchwork of separate tools and databases.

The platform runs locally on Linux or macOS, or via remote machines, and is designed to unify the fragmented software landscape that characterises most pharmaceutical research environments. Anthropic has said it wants first-hand experience using Claude Science on real scientific problems, including drug discovery, so it can improve the product and better understand how R&D workflows actually function in practice.

That’s a candid admission: the internal drug programme is partly a product-testing exercise. But it’s also more than that.

The Coefficient Bio Acquisition

In April 2026, Anthropic acquired Coefficient Bio, a stealth drug-discovery biotech, in an all-stock deal. The acquisition gave Anthropic an in-house team with specialised pharmaceutical planning capabilities and a codebase built specifically for computational drug design. Rather than building that domain expertise from scratch, Anthropic bought it.

Analysts following the deal have suggested the company is aiming to cover the full drug lifecycle — from early-stage discovery through to translation and commercialisation — with potential features including automated pipeline design, synthesis planning and assistance with clinical trial regimen design. Whether Anthropic intends to take any internally discovered drug candidates through clinical trials to regulatory approval remains unclear. Company representatives have indicated the programme is at an early stage, and no specific candidates have been publicly named.

How Long Does Drug Development Actually Take?

The timeline question matters. Drug development from initial discovery to market authorisation typically runs to 10–15 years, according to figures used by regulators including the MHRA and EMA, though the actual duration varies considerably depending on the disease area and development pathway. AI-driven approaches are being explored across the industry to compress those timelines — improving target identification, molecule design and early safety prediction — but independently verified data on how much time or cost AI tools actually save remains limited. Claims of acceleration should be treated with caution until peer-reviewed evidence emerges.

Pharmaceutical and biotech companies are watching Anthropic’s move with mixed feelings. The company presents itself as a potential partner, offering Claude Science as a tool that could optimise their own R&D. But by running its own internal pipeline, Anthropic is also, at least in principle, a competitor. That tension hasn’t gone unacknowledged in industry commentary.

A Broader Shift in AI Strategy

Anthropic isn’t alone in this direction. Several large AI companies and well-funded startups are moving into health and drug discovery, some through licensing tools, others through joint ventures or internal pipelines. The competitive pressure to secure a foothold in high-value sectors like pharmaceuticals is real, and 2026 has seen a notable acceleration of that trend.

But the regulatory environment is unforgiving. Any move from discovery into clinical development would require compliance with frameworks set by the MHRA in the UK, the EMA in Europe, and the FDA in the United States. Those agencies are still developing specific guidance on how AI-generated drug candidates should be validated and documented in regulatory submissions. Researchers in computational biology and pharmacology have broadly welcomed the integration of powerful AI models with bioinformatics workflows, while calling for rigorous peer-reviewed validation before any AI-generated molecule goes near a human trial participant.

Ethics organisations have also raised questions about accountability — specifically, what happens when an AI system contributes to decisions affecting trial participants, and whether proprietary algorithms in public health contexts can be made sufficiently transparent.

What This Means for Kent Residents

Kent patients are unlikely to encounter any direct effects soon, given that drug development timelines run to over a decade and any AI-discovered candidate would need to pass full MHRA review before reaching NHS prescribing. However, if Anthropic’s programme eventually progresses to clinical trials, NHS Kent and Medway ICB and local hospital trusts could become recruitment sites, especially for trials targeting rare or neglected conditions where patient numbers are small and geographically spread. Kent-based universities and life sciences businesses may also find opportunities to engage with platforms like Claude Science as the tools move out of beta and into wider research use.