Dat Tran
Enterprise AI Security Architect
Dat Tran is a Senior Enterprise Security Architect with over eight years working on security platforms in large, complex environments. He currently leads AI security and GenAI governance at a major European utility, where he works on Enterprise AI cybersecurity architecture - including the standards for platforms hosting a large fleet of autonomous and personal agents.
His work sits at the intersection of AI adoption and security engineering: drafting GenAI cybersecurity policies aligned with the EU AI Act, building AI-supported automation for security operations, and leading Zero Trust network and identity initiatives. His background spans network security, cloud platforms (AWS, Azure, GCP), and cloud-native technologies, along with hands-on experience across SIEM, EDR/XDR, firewalls, and microsegmentation.
talks & Q&A
Prompt Injection & the New Attack Surface: A Live Hacking Session on AI Security
Description
Large language models are being wired into everything - agents, copilots, internal tooling - and each integration opens an attack surface that traditional security controls weren't built for. This interactive session puts prompt injection and LLM exploitation on stage, in real time.
Rather than only working through slides, we'll actually break things: direct and indirect prompt injection, data exfiltration through agent tools, jailbreaks that bypass guardrails, and attacks that ride in through documents, web content, and connected systems. For each one, we'll look at why it works, what the model is actually "seeing," and the architectural patterns that hold up against it.
Expect live demos, audience participation, and practical takeaways for your own AI deployments. Whether you're a security practitioner, a developer shipping AI features, or an architect responsible for governance, you'll leave with a clearer picture of how these systems fail and what kind of risks they pose to a wider organization.
