VA’s AI Tools Lack Patient Safety Oversight, Watchdog Warns

The use of AI chatbots by Department of Veterans Affairs (VA) doctors for documenting patient visits and aiding clinical decisions raises concerns due to lack of a formal system to track associated risks to patient safety.
VA’s Office of Inspector General (OIG) identified a potential patient safety risk in how the Veterans Health Administration deploys generative AI chat tools, noting the absence of coordination with the National Center for Patient Safety.
VA GPT, an internal tool, and Microsoft 365 Copilot Chat, a commercial product, both lack a formal mechanism to identify, track, and resolve risks, leading to a missing feedback loop to detect potential safety issues or improve clinical care.
The VA’s AI tools do not have current knowledge bases as they do not access the web, potentially leading to outdated clinical guidance errors.
Despite acknowledging the risks of generative AI, there is no standardized process to manage these risks, which can result in errors that could affect diagnoses and treatment decisions.
The VA’s use of AI is expanding rapidly, with numerous applications in operation, but oversight and guidance on managing AI-related risks remain insufficient.
The OIG’s preliminary review signals urgency for addressing these concerns, as a comprehensive analysis and recommendations will be included in their final report.
The challenges VA faces are mirrored in other federal agencies, with global governments largely lacking centralized AI governance.

You may have missed