Organizing Committee

Industrial KeyNote:

Abstract

Most organizations no longer ask whether to adopt AI. They ask why it is so hard. Gartner expects more than 40% of agentic AI projects to be canceled by the end of 2027, and predicts that 60% of AI projects without AI-ready data will be abandoned. The cause is rarely the model. It is the organization around it, and above all its data.

AI does not just use data; it copies it. One record moves from its source into preparation, training, embeddings, prompts, outputs, logs, and back into training again, and its labels, owners, and access controls do not travel with it. Few organizations have their data ready, and fewer can follow it across that lifecycle.

This talk sets out five harsh realities of industry AI adoption and shows where each one lands on the data. Regulation, compliance, standards, and business integration: there is no finished rulebook, and every obligation attaches to data. Technical depth and alignment: few people understand AI in depth, and fewer understand their own data. Cost, sovereignty, and geopolitics: a provider, a regulator, or a government can switch off the model, and someone else may hold your data. Contingency: AI will fail, can be turned against you, and can be poisoned through its data. Workforce: people are the scarcest part, and almost nobody owns the data end to end.

BIO: William O. Ferguson, an IEEE Senior Member, is Technical Editor of IEEE Std 3536-2026, IEEE Standard for Space System Cybersecurity Design, and leads the Integration Layer subgroup of its working group. His experience spans SES, Thales, and the Space Information Sharing and Analysis Center (Space ISAC).

Acadamic KeyNote:

Abstract:

As Artificial Intelligence (AI) becomes deeply integrated into cybersecurity operations, its dual role as both a defensive enabler and an offensive catalyst is increasingly evident. On one side, AI supports defenders by enhancing detection, prediction, and mitigation of cyber threats. On the other, attackers are swiftly adopting AI models—particularly Large Language Models (LLMs)—to automate and amplify offensive operations.

In this talk, I will explore this evolving landscape and illustrate how AI reshapes both sides of the cybersecurity battlefield. I will present recent advances from our research on applying Reinforcement Learning (RL) for automated attack-path discovery and prediction, as well as on evaluating the offensive potential of general-purpose LLMs for generating vulnerability exploitation steps.

Bio: Abdelkader Lahmadi is a Full Professor of Computer Science at Université de Lorraine in France. He leads the RESIST research group at LORIA and the Inria Centre of Université de Lorraine. His research focuses on bridging the gap between Artificial Intelligence and cybersecurity for complex networked systems, with a particular emphasis on attack detection, prediction, and automated response. Beyond academia, he is the co-founder of CYBI, a cybersecurity company that develops AI-driven attack-path management and prioritization solutions, transferring cutting-edge research into real-world cybersecurity operations.