Tag
This paper systematizes agentic security by evaluating LLM-driven penetration testing tools, identifying recurring failure modes, and deriving quantitative design laws for autonomous security systems.
Microsoft launched its first cybersecurity-specialized AI model, MAI-Cyber-1-Flash, and a new agentic cybersecurity platform called Perception, aiming to compete with Anthropic, Google, and OpenAI in AI-powered security.