From 17 May to 16 June 2026, I had the opportunity to complete a secondment at Montimage as part of the ENSURE-6G MSCA Staff Exchange programme. Coming from LIST, this month gave me the chance to dive into two exciting research directions that sit right at the intersection of AI, explainability, and regulatory compliance.
The main focus of the secondment was exploring the potential of Large Language Models (LLMs) to automate compliance verification of software against the Cyber Resilience Act (CRA). The CRA is reshaping how software products are expected to handle cybersecurity, and the manual effort involved in compliance checking is substantial. Investigating how LLMs could take on some of that burden — and where the real opportunities lie — opened up a genuinely novel research direction, and one we’re already planning to continue building on together.
We also took time to explore how DiCE (Diverse Counterfactual Explanations) could be integrated into Montimage’s open-source MAIP platform, examining how this explainability technique compares with and complements the XAI methods already available in the platform — a topic that feels increasingly relevant as AI systems are deployed in real-world, high-stakes contexts.
What made this secondment particularly valuable wasn’t just the technical output, but the people and the environment. Working in person with the Montimage team allowed for the kind of spontaneous, free-flowing discussions that are hard to replicate remotely. Ideas evolved quickly, questions got answered on the spot, and it was easy to move between exploration and experimentation in a way that felt natural and productive.
The work connects to a broader scientific conversation around applying generative AI to legal and regulatory text analysis — an emerging area where there’s still a lot of ground to cover, and where the collaboration established during this month is expected to lead to further scientific outputs.
I would like to warmly thank Edgardo Montes de Oca, Wissam Mallouli, Manh Dung Nguyen, and Luong Nguyen for their welcome, their time, and the energy they brought to our discussions throughout the month. And of course, thank you to ENSURE-6G for making these exchanges possible — they are a reminder of how much can happen when researchers from different organisations get the chance to simply work together, in the same room, for a little while.
