Before we discuss using AI to find and fix vulnerabilities, we need to first understand some basic AI concepts. This section will focus only on the basics of current AI necessary for our topic.
By the end of this section, you should be able to:
Define core AI concepts including LLMs, AI agents, agent harnesses, and agent skills.
Understand AI capabilities and limitations, including limited context windows, unsound outputs, and susceptibility to prompt injection.
Compare local versus external hosting and evaluate open weights, open source, and closed models.
Examine specialized small language models, safety guardrails, and the history of Cyber Reasoning Systems (CRSs).
Identify key frontier labs, agent frameworks, and security organizations.
Finding and Fixing Vulnerabilities Using AI