Reflections from DataWeek 2025: Sisyphus’ Code – The Ongoing Quest for Trustworthy AI

On the 27th May 2025, the THEMIS project arranged a compelling session at DataWeek 2025, titled Sisyphus’ Code: The Ongoing Quest for Trustworthy AI. Hosted at the OTE Group of Companies (HTO) in Athens, Greece, this workshop brought together experts and stakeholders from across Europe to tackle one of the most persistent challenges in AI development: how to build systems that are truly trustworthy.

Prof Gregoris Mentzas from THEMIS opening the workshop

The Metaphor of Sisyphus – and Trust in AI

Drawing inspiration from the Greek myth of Sisyphus, the workshop title set the tone for a discussion that acknowledged the ongoing, iterative nature of trustworthy AI. Like Sisyphus endlessly pushing his boulder uphill, the effort to align AI with human values, ethical standards, and societal needs is continuous. Each advance in technology presents new risks and dilemmas, requiring persistent re-evaluation and innovation in governance, design, and deployment.

Bridging Research, Regulation, and Real-World Implementation

The session focused on the multidimensional challenge of trustworthy AI, touching on:

• Technical hurdles such as transparency, robustness, and fairness;

• Ethical alignment with societal norms and human values;

• Practical deployment challenges in diverse application domains.

Participants shared perspectives from national, European, and international initiatives, highlighting both progress made and gaps yet to be addressed. This included insights from projects such as Confiance.ai (France), ZERTIFIZIERTE KI (Germany), PHAWM (UK), DeployAI, THEMIS 5.0, FAITH, and broader frameworks like the European Trustworthy AI Foundation and Future-AI.

Voices from Across Europe

Contributions from researchers and practitioners across disciplines and countries enriched the session, including:

• Gregoris Mentzas (National Technical University of Athens)

• Gefion Thuermer, Nicolas Rebierre, Nikos Tachos (FORTH)

• Maximilian Poretschkin (Fraunhofer IAIS)

• Dimitris Apostolou (etami)

• Gabriel Gonzalez-Castañe, Irene Lopez de Vallejo, Julia Palma

Together, they helped spotlight the pressing need for collaborative, cross-sectoral approaches to ensure AI systems are not only powerful, but also trustworthy, accountable, and human-centred.

After DataWeek: The Road Ahead

While no single solution can permanently “solve” the challenge of trustworthy AI, workshops like this demonstrate that meaningful progress is possible. By bringing together technical expertise, ethical insight, and policy experience, the community continues to push the boulder forward—building frameworks, tools, and cultures that prioritize trust in AI.

THEMIS 5.0 remains committed to this mission, actively contributing to and learning from initiatives across Europe and beyond. Stay tuned for more updates on how we’re helping different sectors assess and improve the trustworthiness of their AI systems.