Dr. Thomas Heverin
Director of Technology and Cybersecurity & AI Educator
The Baldwin School
Connect:
About Dr. Thomas Heverin
Director of Technology and Cybersecurity & AI Educator at The Baldwin School. Dr. Heverin has over a decade of experience in cybersecurity, penetration testing, and AI security research. He has published multiple papers with his students at the Baldwin School (an all-girls private school) on topics including prompt injection attacks, ontology-driven cybersecurity, and vulnerabilities in enterprise technologies. He is the author of a Navy cyber risk assessment patent, a NVD CVE entry, exploits on ExploitDB, Google Dorks, and countless bug bounty reports for universities and a government agency.
Talk Details
LLM-SRO: Ontology-Driven Security for Large Language Models
Large Language Models (LLMs) are being adopted across industries, yet their attack surface is expanding faster than defenders can keep pace. This talk introduces LLM-SRO (Large Language Model Security Risk Ontology), an ontology-driven framework for systematically modeling and mitigating adversarial risks in LLMs. Built collaboratively in WebProtégé and paired with AI reasoning through ChatGPT, LLM-SRO integrates the OWASP Top 10 for LLM Applications with MITRE ATLAS adversarial techniques to create a living, queryable knowledge base for defenders. A key takeaway is that LLM-SRO was built with no coding required. This talk equips attendees with practical, actionable methods to prioritize risks and plan defenses.
Session Information
Venue Information
Location:
Live! Casino & Hotel Philadelphia
900 Packer Ave, Philadelphia, PA
Date:
Friday, December 5, 2025