A benchmark for trustworthy clinical AI

A new study published in Nature Communications shows that today's pathology foundation models can be influenced by the origin of a tissue sample. Researchers at BIFOLD and Aignostics developed PathoROB, a first-of-its-kind benchmark to measure and reduce this bias, shaping how the next generation of pathology AI is built.

Photo recap: AI to accelerate scientific understanding

From May 26–29, 2026, ELLIS Unit Berlin and BIFOLD hosted the workshop "AI to accelerate Scientific Understanding" at the Forum Digital Technologies in Berlin, exploring how explaining and interpreting AI models can advance discovery across molecular science, medicine, geoscience, and beyond.

Berlin Summer School of AI and Society

How do you design AI systems that act autonomously while remaining safe, transparent, trustworthy, and accountable? Join the 2026 Berlin Summer School of Artificial Intelligence and Society in Berlin to explore this question. Places are limited to 50. Applications close June 30, 2026.