Podcast
Nov 11, 2024
27 min
English
AI Security – Can We Unpack The Blackbox?
Join top experts live at RSA as they unpack real-world AI security risks, broken standards, and what it takes to secure AI beyond the black box.
Ante Gojsalić
Reet Kaur
Savanah Frisk
John Sotiropoulos
Available on demand
In this live episode of The Elephant in AppSec Podcast, recorded during RSA Conference 2025, industry experts come together to tackle one of the biggest questions in the field: Is AI security a black box problem? Featuring CTO & Co-Founder of SplxAI Ante Gojsalić, former CISO & GRC leader Reet Kaur, AI Safety and Security expert Savanah Frisk, and OWASP GenAI Security Project Co-Lead and Sr. Security Architect at Kainos John Sotiropoulos, the conversation will explore the limitations of current AI security tools and standards – and what it will really take to secure increasingly complex and autonomous AI systems.
The discussion is set to explore why many AI security promises fall short, how organizations can implement an effective AI security posture management, and which vulnerabilities are most critical to address today. With architectures like retrieval-augmented generation (RAG), agentic workflows, and multimodal AI systems on the rise, security teams must rethink how they assess risk and ensure integrity at scale.
Whether you're building AI or securing it, this episode will deliver real-world insights from practitioners, tool builders, and policy shapers working on the front lines of AI security.
What You’ll Learn About Securing AI in the Age of Autonomy
Why AI Red Teaming Needs to Go Beyond the Prompt: Discover how traditional testing methods fall short for agentic AI systems — and why continuous, automated red teaming is essential for identifying real-world exploits like jailbreaks, data leakage, and alignment failures.
How to Strengthen Runtime Security for Dynamic AI Systems: Learn what runtime security looks like in modern GenAI environments, and how to monitor, manage, and defend systems that use real-time decision-making, external tools, and live data sources.
Building a Resilient AI Security Posture Across the Stack: Explore practical strategies for AI security posture management, from policy to implementation — including how to assess and harden systems using RAG, agentic frameworks, and multimodal models.
Resources
Deploy secure and trusted GenAI
