Tag
Anthropic research lead Chris Rohlf summarizes how Claude can automatically analyze and reduce the attack surface of large open-source codebases, running the model over 25 projects and discovering 493 issues to demonstrate practical, economical AI-driven software security improvements.