The Department of Veterans Affairs (VA) reported 367 artificial intelligence (AI) use cases in 2025. AI use case inventoryThis marks a sharp increase from recent years as the department added new applications and retired dozens of older efforts.
VA reported 229 AI use cases in 2024. Notably, VA retired 72 of the 367 use cases in 2025, stating that “development and/or use has since ceased.”
Of the remaining use cases, the ministry reported that 138 have been deployed, 21 are in the pilot stage, and 136 are in the “pre-deployment” stage, or “development or acquisition status.”
“The Veterans Administration is leveraging artificial intelligence to improve how veterans access care, benefits, and services, and to help our employees work more efficiently and effectively,” VA spokesperson Pete Kasperowitz said in an emailed statement to MeriTalk. “The department’s AI inventory demonstrates VA’s commitment to transparency and responsible innovation, allowing us to track, evaluate, and optimize our AI systems while ensuring they are safe, secure, and accountable.”
“This year’s inventory shows steady growth, strengthened governance, and expanded real-world impact, particularly in healthcare, benefits processing, and operational efficiency,” he added.
One of the most prominent use cases that is gaining attention is VA GPT. VA GPT is a generative AI chat interface on the network available to all VA employees to assist with administrative tasks. The VA states that more than 95,000 users have onboarded the product, stating, “Survey results show that more than 70% of users reported increased job satisfaction, and users reported saving an average of two to three hours per week.”
VA is also applying AI to software development. The agency said it has introduced an AI-assisted software development tool that is currently being used by more than 2,000 developers, saving users an average of more than eight hours per week. 74% of users report that this product “helps them focus on more fulfilling tasks.”
Health and medical applications make up a large portion of VA’s AI footprint. Of the use cases introduced and piloted, 107 fall into the health and medical domain.
These include the Veterans Health Administration’s (VHA) Stratification Tool for Opioid Risk Reduction (STORM). It is an AI-assisted clinical decision support system designed to help identify veterans at high risk of overdose or suicide. The VA said use of the system was associated with a 22% reduction in mortality.
Additionally, VA highlighted that the AI-assisted colonoscopy device increased the probability of adenoma detection by 21% and the absolute detection rate by approximately 4% compared to colonoscopy without the device. The higher the adenoma detection rate, the lower the incidence of late-stage cancer and the lower the mortality rate.
in interview Last year, VA Chief Technology Officer Charles Worthington previewed many of these use cases using MeriTalk and predicted that the number of VA’s AI use cases will continue to grow.
“We have a lot of promising use cases, and I think the theme for next year is to expand on some of those most promising use cases,” Worthington said on Meritalk at the time.
The expansion of AI use cases comes as the VA faces increased scrutiny over the use of generative AI in clinical settings. Last month, the VA Office of Inspector General (OIG) issued the following statement: report VHA warns that using generative AI chat tools for clinical care poses “potential patient safety risks.”
The OIG expressed concern about “VHA’s ability to promote and protect patient safety without a standardized process to manage AI-related risks.” However, because the OIG’s review is ongoing, it has not issued a formal recommendation.
“Given the serious nature of the issue, OIG is sharing this preliminary finding broadly so that VHA leadership is aware of this risk to patient safety,” OIG said in the alert.
