Anthropic Is Building Its Own Lab to Cure Diseases Big Pharma Thinks Are Too Poor

The creators of the Claude chatbot are pivoting to molecular biology. Anthropic is utilizing its newly launched Claude Science platform, a tool that aggregates over sixty massive scientific databases, to target neglected tropical diseases like leprosy, dengue fever, and sleeping sickness. These illnesses primarily affect impoverished regions, making them financially unattractive to traditional pharmaceutical giants who prefer funding the tenth consecutive hair-loss drug.

While selling AI subscriptions to biotech firms, the tech startup is now entering the same arena as its own paying customers. The company’s biotech chief, Eric Kauderer-Abrams, defended this double-agent strategy by arguing that developers cannot build effective tools for scientists without getting their own hands dirty in the actual lab trenches. Once the AI identifies viable drug candidates, Anthropic plans to hand them over to traditional pharma companies for actual human testing, though they remain vague about the exact logistics.

This ambitious pivot is legally enabled by the startup's corporate structure as a public benefit company, allowing them to prioritize global health over raw quarterly profits. However, the medical community remains highly skeptical of Silicon Valley's healing touch. No drug designed purely by artificial intelligence has ever successfully cleared clinical trials or received regulatory approval.

Many candidates make it to human trials only to fail spectacularly when confronted with the messy reality of human biology. Frank von Delft, a professor of structural chemical biology at the University of Oxford, pointed out that while these deep learning models are incredible at sorting through data, they are still lightyears away from replacing physical petri dishes and actual wet-lab experiments. Anthropic is now racing against rivals like OpenAI and Google, who are also pouring billions into health tech.

Solving leprosy with code is a magnificent PR stunt, but biology doesn't care about software update cycles. The tech industry's favorite trick of "moving fast and breaking things" takes on a terrifying new meaning when the things being broken are actual human clinical trials.

Source: The Verge