Mina Sadeghi Shaker

Post-deployment changes to large language models can alter behavior while leaving routine outputs largely unchanged, creating a challenge for AI governance when model weights are proprietary. We present a privacy-preserving zk-SNARK-based audit framework that searches for probes designed in the spirit of adversarial examples to amplify logit drift between an approved model and a modified deployment.

α-SMA α-smooth muscle actin ACR American College of Rheumatology CTGF connective tissue growth factor DAPI 4′, 6-diamidino-2-phenylindole dcSSc diffuse cutaneous systemic sclerosis DMEM Dulbecco's Modified Eagle's medium DMSO dimethyl sulfoxide ECM extracellular matrix ET-1 endothelin-1 FBS fetal bovine serum FITC fluorescein isothiocyanate FN1 fibronectin1 FSP fibroblast surface protein GAPDH glyceraldehyde-3-phosphate dehydrogenase IL-6 interleukin-6 lcSSc limited cutaneous systemic...

Introduction The Ras (rat sarcoma virus)-GTPase superfamily consists of GTP-binding proteins with a low molecular weight that convert extracellular signals to various cellular functions. This superfamily is sub-classified into families and subfamilies based on their members’ sequence and functional resemblances.