Where AI assurance ends and AI forensics begins
AI systems are now evaluated before deployment and monitored while they run, but those records were built for detection and prevention, not for investigation after the fact. This session asks whether what assurance leaves behind is enough to reconstruct an AI incident, and what sits in the gap between the two.
RegisterAbout the event
AI assurance has advanced rapidly in recent years. Systems are evaluated before deployment, monitored during operation, and increasingly governed by explicit controls and policies. However, these practices primarily focus on detecting and preventing problems while systems are running.
Forensic investigations face a different challenge: understanding what happened after an incident, often with incomplete evidence and limited access to the original system. Are the records generated through AI evaluation, monitoring, and runtime controls sufficient to support credible investigations and accountability?
The AI Forensics Roundtable brings together leading researchers and practitioners from AI assurance, cybersecurity, and digital forensics to explore where AI assurance ends and forensic investigation begins, and to identify the critical gaps between these fields.
Speakers
Nada Madkour
Director, AI Security Initiative, Center for Long-Term Cybersecurity, UC Berkeley, USA
Panelist
LinkedInRead full bio
Nada Madkour, PhD, is Director of the AI Security Initiative. She specializes in AI standards and policy development, AI transparency and documentation, and AI risk management frameworks. Dr. Madkour plays an active role in critical AI governance bodies, including the NIST AI Consortium, ISO AI standards committees, the EU AI Act Transparency Code of Practice plenary, and the World Economic Forum Safe Systems and Technologies working group. Her expertise spans pioneering AI risk management profiles for agentic AI systems and general-purpose AI models—work that translates technical safety research into practical, implementable governance frameworks for developers and policymakers.
Eoghan Casey
Chief Knowledge Officer, Cyber Knowledge Engineering (CKE); Board of Directors, DFRWS
Panelist
LinkedInRead full bio
Eoghan Casey brings over 30 years of leadership experience in digital forensics, cyber investigations, and the development of investigative standards. As Chief Knowledge Officer at CKE, he works to transform expertise into structured knowledge that improves AI reliability and human performance. He is the author of Digital Evidence and Computer Crime and has contributed to international standards and initiatives, including the Unified Cyber Ontology (UCO), Cyber-investigation Analysis Standard Expression (CASE), and SOLVE-IT.
Brendan Saltaformaggio
Associate Professor, Georgia Institute of Technology, USA; Director, Cyber Forensics Innovation (CyFI) Lab
Panelist
LinkedInRead full bio
Brendan Saltaformaggio is an Associate Professor in Georgia Tech’s Schools of Cybersecurity and Privacy and Electrical and Computer Engineering, where he directs the Cyber Forensics Innovation (CyFI) Lab. His research focuses on cyber forensics, AI security, and investigating untrusted software. His recent work includes techniques for recovering and attributing AI models and LLM conversations from memory. He has led more than $26 million in research funding and received an NSF CAREER Award and best paper awards at leading cybersecurity conferences.
Ali Dehghantanha
Professor and Tier 2 Canada Research Chair in Cybersecurity and Threat Intelligence; Director, Cyber Science Lab, University of Guelph, Canada
Moderator
LinkedInRead full bio
Ali Dehghantanha is a Professor and Canada Research Chair in Cybersecurity and Threat Intelligence at the University of Guelph, where he leads research on AI forensics and the investigation of AI systems. He appeared as an invited expert witness before the House of Commons Standing Committee on Industry and Technology in 2026 and was a Fulbright Canada Visiting Scholar at the University of California, Berkeley.
About the series
When an AI system causes harm, someone eventually has to establish what it did and why. That work usually starts well after the event, often with limited access to the system, and the methods for doing it are still being worked out. Much of the relevant expertise is split between the people who build and assure AI systems and the people who have to investigate them afterwards. Lawyers and regulators are left waiting on the answer.
The AI Forensics Roundtable is a series of short online discussions hosted by the Cyber Science Lab at the University of Guelph. Each edition takes one question from the boundary of the field and brings together a small panel drawn from different sides of it. Sessions are one hour, held online and free to attend. After each one we publish a short synthesis of the discussion.
In partnership with
