Advancing Reliable and Safe AI at MLSS 2026

From 29 June to 3 July 2026, Kraków hosted the Machine Learning Summer School on Reliability & Safety 2026, bringing together researchers, students and AI experts committed to advancing reliable, trustworthy and safe artificial intelligence.

The five-day programme addressed key challenges in modern AI, including safety and reliability, explainability, privacy-preserving machine learning, the security of large language models and emerging research directions. Lectures, practical sessions and discussions allowed participants to engage with recent developments shaping the field.

ELLIS Unit Warsaw contributed to the programme through two lectures. Prof. Wojciech Samek from ELLIS Unit Berlin inaugurated the ELLIS Unit Warsaw Lecture Series with The Science of Understanding AI: From Explanation to Control, dedicated to understanding and managing the behaviour of modern AI systems.

Prof. Tomasz Michalak from ELLIS Unit Warsaw and the IDEAS Research Institute delivered On Stealing Graph Neural Network Models, examining security threats related to graph neural networks and the unauthorised replication of machine learning models.

MLSS 2026 was jointly organised by ML in PL, Jagiellonian University in Kraków, the Group of Machine Learning Research, the Jagiellonian Center for Artificial Intelligence, the IDEAS Research Institute and ELLIS Unit Warsaw.

We thank all speakers, participants, organisers and partners for making the Summer School a valuable forum for scientific exchange and community building around trustworthy AI. The event was held under the Honorary Patronage of the Minister of Digital Affairs of the Republic of Poland.

ORLEN S.A. is the Strategic Partner of the IDEAS Research Institute.

📸: 1, 3, 4: ML in PL; 2: Kacper Cybiński