Anthropic released a vulnerability discovery harness and I want to see if it breaks my code
Anthropic open-sourced a framework for testing how well AI models find security bugs. It includes 32 real CVEs and a scoring system. Time to feed it some of my old projects.
Read the note 2 min read