The Department of the Air Force (DAF) launched an experimental chatbot, NIPRGPT, in mid-2024, aiming to integrate ChatGPT-like capabilities into U.S. military unclassified networks.
Developed by the Air Force Research Laboratory (AFRL) under 'Project Dark Saber,' NIPRGPT was designed to be a secure, military-tailored chatbot, akin to ChatGPT but hosted on government systems.
NIPRGPT allowed authorized personnel (Airmen, Guardians, DoD civilians, contractors) with Common Access Cards (CACs) to interact with AI models for various tasks, such as drafting emails, summarizing documents, and software assistance.
The article's focus is on the security breaches that ultimately led to NIPRGPT being sidelined. While the exact nature of the breaches isn't detailed, the implication is they were significant enough to halt the program.