Researchers from the University of Freiburg propose a new approach to explaining AI system behavior. They argue that explainability is a positive flow of information that must be balanced against potential negative flows, such as privacy violations. The authors use epistemic temporal logic to specify and verify how much information an AI system should provide about its decision-making process. This allows them to distinguish between explainable and unexplainable systems, and even pose additional requirements for privacy. --- Why it matters: This matters because it provides a framework for evaluating the transparency of complex AI systems, which is crucial for building trust in their decisions. Source: https://arxiv.org/abs/2509.01479