With Dr. Madeline Cheah – Associate Technology Director, Cambridge Consultants

‘Agentic AI Security Challenges’ : As we delegate more operational decision-making to machine-driven systems, we expand the threat surface beyond conventional cyberattacks to include AI-specific risks – subtle objective drift, emergent misalignment, model-driven manipulation, and adversarial weaponisation at scale.

This talk examines the security and assurance challenges inherent in granting AI systems real agency, highlights plausible failure modes and rapid-escalation scenarios, and outlines practical strategies for maintaining control, observability, and resilience in autonomous AI deployments.

Bio: Madeline is a cybersecurity and AI assurance specialist with nearly 15 years of experience across autonomous systems, IoT, automotive, and future mobility.

She leads work on AI security testing and autonomous cyber-defence for industry and government clients, with deep expertise in securing complex, opaque, and safety-critical systems.

Madeline has been recognised as one of the Top 100 Great Women in the British Car Industry and an Autocar Driver of Change in Digital Innovation for her contributions to autonomous vehicle security.

She has contributed to platforms including the World Economic Forum and Black Hat USA, appeared in national media, and published multiple peer-reviewed papers. She holds a PhD in automotive cybersecurity, an MSc in forensic computing, and a BSc in biochemistry.