KM news
On September 16, 2026, researchers at the University of Oxford published a study in Nature Machine Intelligence introducing "Epistemic Graph Neural Networks" (EGNN). This framework revolutionizes organizational knowledge management by quantifying the "certainty" and "provenance" of information within a corporate knowledge graph. Unlike standard Large Language Models that often hallucinate, EGNNs utilize a "Probabilistic Logic Layer" to map the reliability of data sources and the strength of logical connections between disparate documents. This allows AI agents to provide not just answers, but a "confidence score" based on the underlying evidence, effectively mitigating the risk of misinformation in high--stakes decision--making environments. The research team announced they will release the "Oxford Knowledge Integrity Toolkit" (OKIT), an open--source implementation of the framework, on November 5, 2026, to help organizations audit their internal AI systems.