March 12, 2026
The Department of Veterans Affairs is expanding its use of generative artificial intelligence across the Veterans Health Administration, including a tool designed to help clinicians document patient visits. As the technology grows, the VA’s Office of Inspector General is taking a closer look at how it is being used and what safeguards are in place.
VA Inspector General Cheryl Mason says her office began reviewing the department’s use of generative AI to better understand how the technology supports clinical care and documentation.
Mason explains that the review was not prompted by allegations or a specific incident. Instead, it is part of the OIG’s broader oversight role as artificial intelligence becomes more common in healthcare.
“We wanted to take a look at how they were using it around clinical care and documentation,” Mason says. “There’s a lot going on in the world of AI. We wanted to understand what they’re doing, how they’re doing it, whether it’s effective and efficient.”
The Veterans Health Administration has been using an ambient AI scribe since late 2023. The generative AI tool listens to conversations between patients and providers and produces clinical notes, helping reduce the administrative workload on clinicians.
According to Mason, the technology operates within a controlled environment inside VA systems. It is not the same as publicly available AI platforms used on the open internet.
“How VA is using it is very condensed to their world,” Mason explains. “It’s very contained in the firewall.”
During the review, the OIG identified a key concern related to the broader characteristics of generative AI. Like many AI systems, these tools can sometimes produce inaccurate or fabricated information—commonly referred to as AI “hallucinations.”
While Mason says the OIG did not identify any patient safety incidents related to the tool, her office found that the department does not yet have a standardized way to report or track AI-related issues within its patient safety systems.
“The primary concern was that there weren’t safeguards beyond the primary processes to report these,” Mason says.
In traditional patient safety processes within the VHA, clinicians can report events such as near misses, safety concerns, or high-impact incidents. However, Mason says the current reporting framework does not include specific categories for AI-related issues.
The OIG issued what Mason describes as a “heads up alert” to highlight the potential risk while the full review continues.
“Our primary role is to recommend that they put something together,” Mason says. “We don’t really tell them what to put together. That’s up to them under their protocols and processes.”
The advisory memorandum raises concerns about the VA’s ability to fully safeguard patient safety without a standardized approach to managing AI-related risks.
At the same time, Mason says VA leadership has been responsive to the concerns raised by the OIG. As discussions continue, the department is already exploring ways to improve how AI-related issues are tracked and addressed.
“They’ve been very responsive during our discussions,” Mason says. “I think they’re already making some changes to address this.”
The OIG is continuing its analysis and expects to release a full report within the next few months. That report will include additional findings and recommendations, as well as acknowledge any improvements the VA has already implemented.
Mason notes that issuing an early advisory like this is relatively uncommon for the OIG, but it allows the department to address potential risks while oversight work is still underway.
“As they make those changes, we’ll acknowledge what they’re doing,” Mason says. “And then when we issue the report, we’ll follow up on any recommendations.”
For Mason, the goal is straightforward: ensure that new technologies supporting veterans’ care are implemented with the right safeguards in place. As the use of artificial intelligence continues to grow across healthcare, oversight and governance will play an important role in maintaining patient safety.