SIG-AI Panel during Security Days 2025

NOTES FOR INTERNAL USE ONLY (TO HELP ON THE ORGANISATION AND MODERATION)

Title:
"AI-Powered Cybersecurity: risks and opportunities"

Wednesday - see programme here.

Timing	Topic & Presenter
5 min	*Moderator explanation of the dynamics of the panel* Presentation of the panelists - Moderator: Albert (SURF)
5 min	Panelist 1 - Jakub Kwiatkowski OR Maksymilian Marcinowski (PSNC) (Eugene will provide their emails)
5 min	Panelist 2 - Martin Zadnik - CESNET
5 min	Panelist 3 - SIG-AI representative (Jan - DFN)
5 min	Panelist 4 - Researcher from Prague, Innovation programme researcher or SURF representative
4 min	*1st question to Panelist 1:*
4 min	*2nd question to panelist 2:*
4 min	*3rd question to panelist 3:*
4 min	*4th question to panelist 4:*
-- min	Open questions from audience AND prepare some backup questions to stimulate discussion
10 min	Closing phrases from each panelist and final acknowledgments from the moderator

Suggested questions (to be discussed):

What is the mayor challenge for NRENs to use AI?
What is the mayor risk of using AI for security?
Will AI-based threats lead to a major change in cybersecurity?
What is the most important impact/contribution of LLMs in cybersecurity?
Who profits more red team or blue team? (are the models more beneficial to attackers or defenders?)
Is the availablibility of data sufficient for cybersecurity? What prevents of having the datasets available? To avoid bias and to have a full trained model, is important to have data from other NRENs too.
How can we increase the trust of network operators in automate decision based on ML models in cybersecurity (ex: with active decisions of blocking something)?
What are the important aspects to consider to implement AI in cybersecurity ( legal, ethical, others)?
How to provision/have - compute capacity for NRENs to train ML models for cybersecurity?
How can AI be related with emerging technologies (IoT, Quantum, 5G/6G, blockchain)?

Space shortcuts