In a blog post, the EF outlined how AI agents are beneficial in finding vulnerabilities, but most turn out to be false positives.While AI agents help scale vulnerability testing, they also require humans to review more findings to determine which bugs are real.