LLMs and particularly Claude are very capable security engineers. My startup builds offensive pentesting agents (so more like red teaming), and if you give it a few hours to churn on an endpoint it will find all sorts of wacky things a human won't bother to check.
No comments yet.