Agentic Security Risks – John Sotiropoulos – OWSP25 #1
John Sotiropoulos is the Head of AI Security Kainos helping safeguard national-scale projects in government, healthcare, and finance. He serves on the OWASP GenAI Security and Top 10 for LLMs Project Board of Directors and co-leads the OWASP Agentic Security Initiative , driving the development of the forthcoming Top 10 for Agentic Applications. John has authored the UK Government’s official Implementation Guide for the AI Cyber Security Code of Practice, now adopted as a global ETSI standard and his book on Adversarial AI is an Amazon bestseller. He leads the national AI Security conversation in the UK and is the founder and chair of the new AI Cyber UK Network.
GenAI Security From Assessment to Action – Sandy Dunn – OWSP25 #1
- AI Threat Defense COMPASS ties guidance from the other OWASP GenAI project together so leaders, developers, AI experts, and cybersecurity experts can operationalize the information.
- As GenAI initiative teams we are able to work together which accelerates securing GenAI at the Speed of AI: Delivering information on what's needed now, with agile iterations as apps and infrastructure evolve with actionable guidance.
Sandy Dunn is a CISO with over two decades of experience spanning manufacturing, healthcare, and high-growth startups. As CISO at SPLX.AI, she leads the security strategy for the company’s automated and continuous AI Security and Red Teaming platform built to defend Conversational and Agentic AI systems at scale.
Sandy is a core contributor to the OWASP GenAI Project and serves as the creator and project lead for both the OWASP GenAI Cybersecurity & Governance Checklist and the OWASP GenAI Threat Defense COMPASS
In addition to her industry leadership, Sandy is an Adjunct Professor at Boise State University, where she teaches cybersecurity courses and mentors the next generation of security professionals. Her expertise spans enterprise security architecture, AI risk governance, red teaming methodologies, and the integration of AI-specific threat modeling into modern security programs.
