A recent report by the VA’s Office of the Inspector General (OIG) found that the U.S. Department of Veterans Affairs (VA) deployed generative AI chat tools for clinical staff without sufficient oversight or safeguards, creating potential risks to patient safety.

In a review conducted between October 2025 and February 2026, the OIG found that the Veterans Health Administration (VHA) was actively using tools such as VA GPT and Copilot Chat in clinical settings despite lacking governance measures needed to monitor errors and manage risks.

In contrast, the VA’s pilot of Ambient AI Scribe, which assists with drafting clinical notes, included stronger oversight mechanisms.

The report expressed concern that efforts to reduce bureaucratic oversight, in line with the Office of Management and Budget’s 2025 AI guidance, may have weakened risk management practices. Investigators found that VA leaders prioritized expanding access to AI tools and viewed some safeguards as impractical.

The report also criticized limited coordination between AI leaders and patient safety officials, noting that generative AI was at times treated similarly to traditional search engines despite its ability to generate, synthesize and present information.

The OIG recommended classifying AI chat tools as high-impact systems, strengthening governance, implementing additional safeguards and integrating AI-related risks into patient safety programs. The VA agreed with the recommendations and outlined plans to improve AI governance, monitoring and staff education.