The SnT (Interdisciplinary Centre for Security, Reliability and Trust) is rock ’n’ roll and no longer really needs to prove it. But AC/DC blasted out ‘Thunderstruck’ to set the scene for the main event of this year’s Partnership Day, held on Thursday 21 May at the conference centre. At the risk of bursting the audience’s eardrums. The messages, however, were worth hearing. Behind the atmosphere of a full-blown tech show, the research centre affiliated with the University of Luxembourg wanted above all to get one idea across: artificial intelligence (AI) is no longer an experimental subject, but an industrial, economic and now strategic issue for Europe.

With nearly 850 participants this year, compared to around 50 at the first few events, the SnT Partnership Day has become one of Luxembourg’s leading events focusing on applied research, innovation and technology transfer. “We are all here today because we know that we are living through a period of truly momentous change,” summarised the Rector of the University of Luxembourg, Jens Kreisel, referring to a pace of technological change that has become difficult to keep up with, even for specialists. “Watching a tsunami roll in is not a strategy,” he added, arguing that businesses, institutions and researchers must now work together to prepare for the massive impact of AI.

The theme of this year’s event, “Building Trust in AI”, served as the central theme for all the presentations. The director of SnT, Yves Le Traon, spoke at length about the very difficulty of defining this concept of trust as it applies to artificial systems. “Trust is based on a relationship,” he pointed out, before comparing the relationship between humans and AI to that with a dog: a deliberately provocative parallel designed to illustrate the emotional, irrational and sometimes deceptive aspects of technological trust. In his view, AI can produce convincing responses whilst being spectacularly wrong, sometimes in ways that are far more difficult to detect than a typical human error.

A matter of trust

The SnT’s key message is clear: trust in AI can no longer be based on subjective impressions or the reputation of models alone. It must become measurable, objectifiable and subject to continuous testing. Yves Le Traon outlined the work being carried out on bias, system robustness, deepfakes, model explainability and regulatory compliance. The SnT is also preparing to launch an “AI Sandbox”, developed in collaboration with the List and LuxProvide and integrated into Luxembourg’s AI Factory, to enable companies to test and assess in practical terms the level of trust they can place in their AI systems.

We’re really shooting ourselves in the foot.

Carlo Duprel, head of technology transfer office, Science and Technology

Beyond technology, the contributions also revealed a growing sense of concern across Europe. Charles Duprel, head of technology transfer at SnT, delivered a particularly political plea regarding Europe’s lag in innovation. In his view, Europe suffers less from a lack of research than from a structural inability to turn its scientific discoveries into rapid commercial innovations. “In Europe, we make bottle caps attached to plastic bottles,” he quipped, before acknowledging that this regulatory approach remains “reasonable” and consistent with European environmental priorities.

But behind this quip lies a much deeper critique of how the EU operates. Carlo Duprel believes that state aid rules and regulatory mechanisms significantly complicate the transfer of technology between universities and businesses. “We are really shooting ourselves in the foot,” he said, condemning decades of administrative complexity and rules that are sometimes impossible to understand when it comes to commercialising public research.

€150bn to invest… and little return

The issue of technological sovereignty was raised in virtually every speech. Yves Le Traon emphasised the need for Europe to develop sovereign technologies in the fields of AI, secure communications, autonomous systems and cybersecurity. The partnership with Mistral AI, which has become a European flagship for generative AI in the face of competition from American and Chinese giants, was highlighted at length.

Guillaume Bour, vice-president of Mistral AI, pointed out that his group now has nearly 1,000 employees less than three years after its creation and already has a presence in Luxembourg. He also offered a rather blunt assessment of the European AI market. According to him, nearly €150bn is expected to be invested in AI in Europe over the coming years, whilst “70% of AI projects fail to deliver value”. Among the causes identified are regulatory fragmentation across Europe, a shortage of talent, data quality and the difficulties companies face in maintaining control over their systems.

This technology is far too powerful to remain in the hands of just a few very large companies.

Guillaume Bour, vice-president, Mistral AI

Mistral AI advocates a European approach based on open source, flexible deployment and model customisation. “This technology is far too powerful to remain in the hands of just a few very large companies,” insisted Guillaume Bour, implicitly referring to the American and Chinese groups that currently dominate the global market.

Towards a future “national champion”

The minister for Research, Higher Education and Digitalisation, Stéphanie Obertin, also focused her remarks largely on the issues of sovereignty and economic impact. Several key priorities clearly emerged: accelerating the translation of research into industrial applications, strengthening the links between research and the economy, and positioning the SnT as the future ‘national champion of AI’ – a role it already fulfils, judging by its statistics: 822 publications, 43% of which are in the top 10, over 360 completed PhDs, six ERC Grants, five FNR Pearl Chairs and recognition from a heavyweight, Google, which opened its centre of excellence in Luxembourg in February 2025.

The minister also emphasised the need to develop technologies “that we can control ourselves and share with our trusted partners”. In this respect, the SnT Partnership Day is the perfect showcase for the gradual shift in tone in Luxembourg. AI is no longer presented solely as an academic subject or a mere economic opportunity. It is also becoming an instrument of power, competitiveness and sovereignty in a context of increasingly overt global technological rivalry. Behind the robot demonstrations, the technology stands… and the references to AC/DC, the political message was ultimately much more serious than it appeared.