AI systems are entering war zones faster than anyone is testing them for it.

​Conflicts are fragile settings involving violence, adversarial actors, a dedicated legal framework (International Humanitarian Law/Law of Armed Conflict), and harms – humanitarian consequences, including death and displacement. Yet the testing standards of AI systems are lagging their deployment in such high-stakes environments.

​The closed event will bring AI evaluators and humanitarian practitioners together to explore how the combined expertise can help close that gap – towards safer and more effective products.

​This event is part of Geneva Loading…, a multi chapter program preparing eleven Swiss-based thought leaders from academia, civil society, media and policy to contribute to the AI Summit 2027 in Geneva. The program is a collaboration between Swissnex in San Francisco and Mercator Foundation Switzerland, supported by Gesda, AlgorithmWatch, and TA Swiss.

​*Who should attend? AI safety and evaluations specialists interested in an open exchange on jointly improving AI safety in conflict settings, including through better evaluations, are invited to apply. Seats are very limited.