In most penetration tests targeting AI agents, we see a pattern: Across different organisations and tech stacks, the same types of vulnerabilities keep appearing. And, surprisingly often, this affects two layers that are underestimated or completely overlooked.
This raises the question: why does this happen so consistently? And what does this mean for the security of your own agents?
In this webinar, you will learn:
- Which typical findings recur time and again in AI agent systems
- Which two layers are particularly critical
- How you can check your systems using a concise checklist
Register and find out just how secure your AI agents really are.
- Date: Thursday, 21.05.2026
- Time: 11:15 – 12:00
- Language: German
Registration Agentic AI Security Webinar
Please check your spam folder if you do not receive the confirmation email after registering.
About the Speaker
Dr. Marco Bertenghi earned his doctoral degree from the Zurich Graduate School in Mathematics (ZGSM, UZH/ETHZ) in the field of mathematics, specializing in memory and diffusion processes, inspired by the dynamics of artificial intelligence (AI) systems. Since then, he has seamlessly applied this solid analytical and mathematical
foundation to the field of cybersecurity.
Since 2023, Marco Bertenghi has been working as a penetration tester (since 2026 Senior) at Oneconsult AG, where he actively contributes to strengthening companies’ cyber resilience through security testing. He has already led several projects focused on AI security – including in the insurance industry, the education sector, and the banking industry.
