Anthropic 利用 Claude 进军药物研发

Anthropic wants to make its own drugs with help from Claude.

Claude Science is meant to do for lab research what Claude Code did for programming.

Drug discovery is messier, because molecules fail when biology, chemistry, safety, or dosing breaks.

Anthropic wants its researchers to feel those failures directly while shaping the product.

The first programs will target neglected diseases, including rare disorders and some tropical conditions.

These areas often have clearer biology but weaker commercial incentives than common diseases.

Many rare genetic disorders start with one broken gene, giving AI a cleaner target.

Common conditions usually involve many genes, tissues, histories, and environments interacting at once.

That difference makes early rare-disease work a harder business, but a clearer testbed.

Anthropic also avoids competing with pharma customers by choosing overlooked programs.

Anthropic already works closely with major drugmakers, so it understands where researchers get stuck.

Novartis CEO Vas Narasimhan joining the board gives those product decisions direct pharma-level pressure.

Anthropic bought Coefficient Bio and hired AlphaFold researcher John Jumper, so this is not a side experiment. Shows how seriously Anthropic is treating biology.

---

statnews. com/2026/06/30/anthropic-ai-drug-development/