The University of Vermont will coordinate a $19 million, five-year National Science Foundation project to design and test a national observatory for exploring life’s smallest structures in three dimensions. Working with UC San Diego, researchers will combine advanced electron microscopy, artificial intelligence and powerful computing to expand access to imaging capabilities currently concentrated in specialized laboratories. Pilot studies will investigate how learning changes brain wiring, how vision varies across species, how a parasite crosses the blood-brain barrier, and how single-celled organisms are organized. The project will lay the groundwork for a shared national resource serving researchers across biology and medicine.

As autonomous AI agents evolve from conversational assistants into true scientific collaborators, the pharmaceutical industry is entering the next chapter of its AI transformation, one in which discovery workflows are not merely assisted by AI, but orchestrated by it. To help accelerate the advent of Pharmaceutical Superintelligence, Insilico Medicine (03696.HK), a clinical-stage generative AI-driven drug discovery company, today announced that the Pharma.AI Webinar 2026 Q3 Fall Updates will take place on September 30, from 10 EST. Registration and event details are available at: https://insilico.zoom.us/webinar/register/2417878444165/WN_gLMYTGF2RYW3EoupiucfTA#/registration

A new artificial intelligence (AI) framework that combines ultrasound (US) with digital breast tomosynthesis (DBT) may improve breast-level risk classification by raising specificity while preserving sensitivity. 

A candid debate on AI in research, a warning about quantum-era data security and the Forum's first participant-driven "Un-Conference" defined a week that brought 21 laureates together with 200 young researchers from more than 50 countries

A multicenter study of 2,746 patients with acute pancreatitis (AP) introduces ORACLE, an artificial intelligence model that predicts organ failure from CT imaging. The model achieved AUCs of 0.81–0.89, outperforming conventional scoring systems, and provided a median early warning of 3.5 hours before clinical onset. Its 97.2% negative predictive value may help clinicians stratify risk.A multicenter study of 2,746 patients with acute pancreatitis (AP) introduces ORACLE, an artificial intelligence model that predicts organ failure from CT imaging. The model achieved AUCs of 0.81–0.89, outperforming conventional scoring systems, and provided a median early warning of 3.5 hours before clinical onset. Its 97.2% negative predictive value may help clinicians stratify risk.