On April 14, the Security Network Munich community gathered at KPMG’s Ignition Center in Munich for the April edition of Cybersecurity Meetup Munich. Hosted together with KPMG, the evening focused on AI Security and brought together cybersecurity professionals, technology experts, and members of Munich’s innovation ecosystem for an evening of expert insights and discussion.
After a short welcome by the organizers and hosts, the evening continued with two talks approaching AI Security from different but highly complementary perspectives.
First Session – “Security in the Age of Computer Use Agents – with Live Demonstration” Presented by Markus Hupfauer, Solution-Lead AI Security at KPMG FS Technology & IT-Compliance
In the first session, Markus Hupfauer explored how AI is evolving from passive chatbots into autonomous computer-use agents capable of interacting with interfaces and acting on behalf of users – significantly expanding the potential attack surface.
A key focus of his talk was the growing relevance of prompt injection and permission abuse. Through a live demonstration, Markus showed how malicious instructions can be embedded in seemingly harmless content, leading agents to act with the user’s own privileges. This highlighted a critical point: traditional guardrails alone are not sufficient.
One of the central takeaways was that AI security ultimately breaks at permissions. Instead of assuming full control over agents, organizations need to design systems with failure in mind. Approaches such as Zero Trust, strict isolation, and minimizing blast radius become essential when treating AI agents as inherently untrusted.
Second Session – “AI Security Beyond the Hype – From Buzzwords to Empowerment” Presented by Shuying Li, Assistant Manager, Cyber Security & Resilience at KPMG
The second talk took a broader perspective, focusing on how AI is reshaping the threat landscape from an organizational point of view.
She emphasized that AI does not introduce entirely new threats, but rather amplifies the speed, scale, and complexity of existing ones. At the same time, many organizations face not just technical challenges, but gaps in understanding and decision-making.
Shuying framed AI Security as an organizational capability rather than a purely technical function. Her talk highlighted the importance of clear responsibility, practical guardrails, and human judgment, alongside technical controls. She concluded with practical ideas for enabling teams to use AI safely and confidently, turning uncertainty into resilience rather than fear.
Taken together, both talks showed that AI Security is no longer a future topic. As organizations continue to adopt AI across more functions and processes, the need for practical security measures, sound governance, and organizational readiness continues to grow.
Following the presentations, attendees stayed for networking and further discussion, using the opportunity to exchange perspectives on current developments, practical challenges, and the wider implications of AI for cybersecurity. As always, these conversations were an important part of the evening and reflected the value of the Meetup format for the local community.
A big thank you to KPMG for hosting this edition of Cybersecurity Meetup Munich, to Markus Hupfauer and Shuying Li for sharing their insights, and to everyone who joined us for another engaging evening in Munich. The next edition promises to be equally compelling – we’re already looking forward to the May edition!