How to support AI project adoption at Les Rencontres de la Sécurité 2025
Les Rencontres de la Sécurité (11th edition), Casino 2000, Mondorf-les-Bains, Luxembourg, June 11, 2025, with Johann Alessandroni
At the 11th edition of Les Rencontres de la Sécurité, the annual cybersecurity event organised by Thales in Luxembourg, I gave a talk with Johann Alessandroni, Information Security Governance Team Leader at Thales, titled “How to support AI project adoption: competences, governance and security principles”.
AI is moving from experimentation to real-world deployment, driven by generative AI, while policies, skills and security controls often lag behind actual usage. The talk covered:
- Where organisations stand: shadow AI, missing AI policies, the first AI-related security incidents, and data quality and governance as the main obstacles to adoption
- The pain points: no clear AI governance or oversight, weak data governance, a limited understanding of AI-specific risks, and security capabilities that do not yet cover AI systems
- Three streams to manage AI systems: education and awareness, AI governance and risk management (AI Act, GDPR), and the secure deployment and monitoring of AI systems
- A secure AI journey in five steps: education, definition of use cases and roadmap, governance and risk management, implementation of a secure framework, then run and evolve, including offensive testing
- From discovery to run: how organisations use, build or provide AI, the most common use cases we see, and the security work along the way, from spotting shadow AI and preventing data leaks to securing AI applications and the development lifecycle, auditing, and red teaming AI systems
A year later, I came back to the 12th edition with NeoVoyage, a talk on agentic AI security.
The visual at the top of this page is the talk’s title slide.