AI security in real life — Thales EU User Group Conference 2026

Thales EU User Group Conference (fourth edition) — Brussels, February 5, 2026

At the 2026 Thales EU User Group Conference, over 40 representatives from EU Institutions, Agencies and Bodies — including the European Commission, ESA, EEAS, EDA, EUMETSAT, EUROCONTROL, ECSO and the Council of the European Union — gathered in Brussels for high-level discussions on the digital topics shaping Europe’s future: EU sovereignty, cyber resilience, protection of critical systems, AI and sensitive data, and the transition to post-quantum cryptography.

I gave a talk on AI security — AI security in real life: not a buzzword, but a concrete operational challenge for European organisations. Building on a real-world chatbot scenario, I showed how an AI model can leak personal data, how to build secure architectures, and why aligning with the EU AI Act will become a defining capability for all regulated organisations. Right after the talk, I sat down for a video interview summarizing these insights.

Links #

• LinkedIn post and interview by Thales Cybersecurity Services
• LinkedIn post by Lydia Grilheres (Thales)

Archived copy #

From the LinkedIn post by Thales Cybersecurity Services (archived July 2026):

At the 2026 Thales EU User Group Conference in Brussels, over 40 representatives from EU Institutions, Agencies and Bodies came together to engage in high level discussions on the digital topics shaping Europe’s future.

One key highlight: AI security in real life, not as a buzzword, but as a concrete operational challenge for European organisations. To bring this to life we sat down with Julien Ehrhart, AI Security Advisor just after his presentation.

He shared powerful insights based on a real world chatbot scenario including how an AI model can leak personal data, how to build secure architectures, and why aligning with the EU AI Act will become a defining capability for all regulated organisations.

Interview video from the LinkedIn post (local archive).