Securing Agentic AI with the NIST Cybersecurity Framework 2.0

Agentic AI —LLM-powered autonomous agents— is reshaping cybersecurity paradigms, introducing a novel attack surface that exposes gaps in current security approaches. This Systematic Literature Review (SLR) of 30 peer-reviewed papers examines the emerging threats and mitigation strategies for these agentic systems. The review synthesizes evidence through a four-dimensional taxonomy derived from the OWASP Agentic AI Threats framework. Building on these findings, the paper proposes a new adaptation of the NIST Cybersecurity Framework (CSF) 2.0 to guide organizations in identifying, protecting, responding to, and recovering from risks associated with agentic AI. The presented framework provides a clear and practical method for securing agentic AI-driven systems in both enterprise and research contexts.