The brief · 20 February 2026

The AI brief, 2026-02-20

A lab is betting on AI to dream up new molecules while Anthropic is teaching Claude Code to hunt down real security flaws. Two ways of using AI to spot what's broken before anyone else does.

1 min read J / K to navigate

Insilico's pitch: a single prompt that invents a drug candidate

Insilico Medicine just published its vision for pharma R&D run end-to-end by AI agents, from target discovery to clinical trial design. If you're just using ChatGPT day to day, this feels far removed from your world. But it signals where prompt-driven reasoning is headed: not just writing text, but orchestrating an entire lab.

practitioners › Perspective published February 20, 2026 in ACS Central Science, citing the use case: "Design a drug for idiopathic pulmonary fibrosis."

insilico.com →

Claude Code Can Now Hunt Down Your Security Vulnerabilities

Anthropic is building a vulnerability scanner and patch generator directly into Claude Code. It's Enterprise-only for now, but this is a preview of what's coming to your IDE in the next few months.

practitioners › Preview limited to Enterprise and Team accounts, no public access or pricing announced yet.

Anthropic →