AI & Security Webinar – Presentations and Recording Now Available

 

All presentations from the AI & Security Webinar are now available on Zenodo, and the full session recording can be watched on YouTube. The webinar, held online on 30 April 2025, was jointly organized by ELASTIC, PREDICT-6G, HARPOCRATES, CONFIDENTIAL6G, and RIGOUROUS, with support from FAITH, CUSTODES, and 6G-Cloud.

The event explored how artificial intelligence and cybersecurity intersect in next-generation networks, with sessions covering privacy-preserving AI, AI for resilient networking, and security and privacy in orchestration.

Talks and Summaries

  • Cybersecurity Certification vs. Trustworthiness Certification for AI Systems — Overview of certification approaches, comparing security compliance with broader trustworthiness assessment in AI.

  • Efficient Privacy-Preserving Feature Selection — Method for selecting relevant features for AI models while keeping sensitive data protected.

  • Privacy-Preserving Hyperparameter Tuning for Federated Learning — Technique to optimize federated models without exposing raw data or sensitive parameters.

  • Advancing Robust and Privacy-Enhanced Federated Learning in 6G Networks — Enhancements to federated learning to improve robustness and privacy in distributed 6G environments.

  • AI/ML for Incident Detection and Mitigation in 6G Networks — Machine learning methods for detecting and responding to network incidents in real time.

  • Enabling Decentralized Spectrum Access with Blockchain — Use of blockchain to manage fair and secure access to wireless spectrum.

  • DecSecOps Deployment of Security & Privacy in Multi-Segment Networks — Integrating DevSecOps principles to ensure privacy and security in complex multi-segment network deployments.

  • AI-Driven Self-Healing Edge/Cloud Operations for Energy Saving in 6G Networks — AI-based orchestration to optimize operations and reduce energy usage in cloud and edge systems.

  • Resilient Cyber-Physical Devices with WebAssembly Sandboxing — Applying WebAssembly for secure, sandboxed execution in connected devices to improve resilience.

Access the materials