TITLE: Pentagon Replaces Air Force AI Chatbot with Advanced Tool

Pentagon Phases Out Air Force's Popular AI Chatbot

The U.S. Air Force is shutting down its experimental artificial intelligence chatbot, NIPRGPT, effective New Year's Eve, replacing it with a more powerful Pentagon-wide system designed to accelerate military adoption of generative AI.

NIPRGPT, short for Non-Classified Internet Protocol Router Generative Pre-Trained Transformer, was developed by the Air Force Research Laboratory and launched in June 2024. It allowed personnel to experiment with generative AI for tasks like research, writing memos and generating code, while incorporating safeguards to ensure responsible use.

Air Force officials introduced the tool before full perfection to meet the military's need for rapid technological integration. Within three months, more than 80,000 service members used it, and overall, over 700,000 people across the Defense Department accessed the chatbot.

"Technology is learned by doing," said Chandra Donelson, chief artificial intelligence officer for the U.S. Space Force, a branch of the Air Force. She noted that frontline personnel's experiences are informing future policies, acquisitions and investments.

The discontinuation stems from the rollout of GenAI.mil, a new generative AI platform intended for all armed forces branches, serving nearly 3 million users. Hosted on Google's Gemini for Government environment, unveiled in August, GenAI.mil is described as significantly more capable than NIPRGPT.

A Defense Department release stated that GenAI.mil will enable an "AI-first" workforce, improving efficiency and readiness. It is expected to transform intelligence gathering, operational planning, logistics and other functions, fostering a cultural shift toward AI-driven operations on the digital battlefield.

Defense Secretary Pete Hegseth emphasized the initiative's strategic weight, likening it to "pushing all of our chips in on artificial intelligence as a fighting force."

Military AI Push Amid Global Challenges

The transition reflects the Pentagon's broader effort to counter evolving threats and maintain U.S. military superiority. Emil Michael, undersecretary of defense for research and engineering and a former executive at Uber and other tech firms, highlighted China's rapid military expansion as the most significant buildup in modern history over the past 10 to 15 years.

Michael pointed to the war in Ukraine as a key influence, where low-cost robots and drones are increasingly dominating the battlefield, challenging the U.S. reliance on high-end, expensive systems. "We have to start shifting the needle from all exquisite systems that are built in expensive ways to a more modern architecture," he said at a recent gathering of military journalists.

The military has already integrated AI in various ways, including enhanced simulators for pilot training at bases like Laughlin Air Force Base near Del Rio, Texas, and virtual reality headsets for Navy training courses.

GenAI.mil's deployment is part of a larger strategy to leverage AI for competitive advantage. By centralizing access across services, the system aims to streamline experimentation and deployment, ensuring the U.S. military keeps pace in the global AI race.

Air Force leaders initially prioritized speed over perfection with NIPRGPT to equip personnel with advanced tools "at the speed of relevance." Its success demonstrated the demand for such capabilities, paving the way for broader implementation.

As the Pentagon advances, officials stress the importance of safeguards to mitigate risks associated with generative AI, including data security and ethical use in non-classified environments.

The shift from NIPRGPT to GenAI.mil underscores a rare instance of one AI tool displacing another, mirroring broader concerns about automation's impact on jobs — but in this case, within the realm of technology itself.