WASHINGTON — United Launch Alliance (ULA) is testing an innovative artificial intelligence system called ‘RocketGPT’, specifically designed for government and defense users. This marks one of the initial deployments of an AI chatbot built to meet stringent security standards required in the defense sector.
The system operates on Microsoft Corp.’s Azure cloud platform, which is approved for handling sensitive government data under International Traffic in Arms Regulations (ITAR)—a comprehensive set of security standards governing aerospace and defense information.
Approximately 150 ULA employees have trialed ‘RocketGPT’, which aims to assist with routine tasks such as generating reports, drafting government proposals, and analyzing flight telemetry data. ULA CEO Tory Bruno expressed enthusiasm about the system, stating that ‘we’re super excited’ and highlighting its potential to ease tedious workloads.
The development of ‘RocketGPT’ highlights a significant step in AI adoption within the defense industry, where consumer-grade AI models are typically prohibited due to security concerns. Instead, contractors are opting for solutions built on officially approved platforms tailored for sensitive data handling.
Security and compliance are top priorities
To customize the chatbot for its specific needs, ULA collaborated with OpenAI and PwC, dedicating a team of about 20 personnel. ULA emphasizes that this AI system is intended to support, not replace, human workers—addressing concerns about job displacement. Bruno explained, ‘AI is really, really good at handling massive volumes of data,’ adding that it’s intended to enhance productivity rather than eliminate roles.
OpenAI’s efforts with ULA represent a broader push into government markets. Recently, OpenAI secured a Pentagon contract worth $200 million over a year to develop AI capabilities tailored for national security applications.
Cautious optimism
Bruno cautioned against overestimating AI’s current capabilities, warning that AI systems are not plug-and-play solutions. ‘You don’t just grab an AI off the shelf and just turn it on,’ he noted. ‘It requires extensive training on large datasets.’ He also emphasized that human oversight remains essential, with employees responsible for final decisions and outputs.
Despite modest investment estimates, the pilot aims to improve efficiency and output quality across departments including engineering, legal, and finance. The goal is to explore broader applications of AI within ULA’s operations.
As the defense sector innovates, this deployment demonstrates a cautious yet progressive approach to integrating AI technologies under secure and compliant conditions. ULA’s initiative could serve as a template for future government and defense applications of artificial intelligence.
