Webinar Description
Key Takeaways
- Live demonstrations recreate real-world AI escape incidents to show how agents attempt to bypass security controls
- Covers testing methodologies for AI containment, permissions and tool access under adversarial conditions
- Introduces the Cloud Range AI Readiness Framework with five steps to assess AI agent trustworthiness
- Designed for CISOs, security operations leaders and AI engineering professionals in enterprise, government and critical infrastructure
- References industry standards including MITRE ATLAS and NIST AI RMF
About the Webinar
Putting AI Escape Techniques to the Test is a live virtual webinar examining security challenges that emerge when advanced AI agents operate outside their intended boundaries. The session features speakers from Cloud Range and Fire Mountain Labs who will demonstrate AI escape behavior through practical recreations of documented incidents.
The webinar moves beyond theoretical discussion to show attendees how AI agents respond when confronted with security controls. Demonstrations will reveal the techniques agents use when attempting to circumvent restrictions, providing insight into AI persistence patterns that security teams need to understand when deploying autonomous systems.
AI Containment and Adversarial Testing
A central focus of the session is testing AI containment strategies under realistic adversarial conditions. The webinar addresses how organizations can validate that their AI agents remain within defined operational boundaries, covering technical controls around permissions and tool access alongside operational considerations for human oversight.
The session introduces the Cloud Range AI Readiness Framework, a structured approach comprising five steps for assessing and demonstrating AI readiness. This framework aims to help organizations establish measurable trust in their AI systems rather than relying on assumptions about agent behavior.
Standards and Governance Context
The webinar content aligns with established AI security standards, referencing both MITRE ATLAS (Adversarial Threat Landscape for AI Systems) and the NIST AI Risk Management Framework. These frameworks provide the broader governance context for the containment and testing methodologies discussed during the session.
Who Should Attend
The webinar targets security and AI leaders responsible for deploying or governing autonomous agents. Relevant roles include CISOs, security operations leaders, AI and ML engineering leaders overseeing agent deployment, and professionals working in AI governance and risk management. The content is particularly applicable to those in enterprise, government and critical infrastructure sectors where AI agent security carries significant operational and compliance implications.