Why It Matters

The Department of Veterans Affairs (VA) administers one of the federal government's largest disability benefit programs, distributing over $195 billion to over 6.9 million veterans and their families in fiscal year 2025. On July 10, Government Accountability Office (GAO) Chief Scientist Sterling Thomas testified before the House Committee on Veterans' Affairs, Subcommittee on Technology Modernization, summarizing GAO's body of work from September 2015 to June 2026 on VA disability compensation claims processing, IT modernization, and AI. The testimony is not a standalone investigation but a synthesis of prior GAO findings delivered as Congress scrutinizes VA's push to automate claims decisions.

GAO found gaps in VA's oversight of exam quality from contracted medical providers and in the management of training for claims processors. VA has taken some steps to address 15 remaining GAO recommendations but has not fully implemented them. The agency is simultaneously exploring AI to further automate claims processing, a move GAO flagged as risky given unresolved management weaknesses.

The Big Picture

The Honoring our PACT Act of 2022 required VA to develop a plan to increase both the speed and accuracy of claims processing decisions. Separately, VA launched the Veterans Benefits Management System in 2009 to build an electronic, paperless claims infrastructure. GAO found that effort was not driven by robust planning and did not include goals for system response times or user satisfaction, making it difficult to measure whether modernization was actually improving service to veterans.

Without clear performance benchmarks, there is no reliable way to determine whether veterans are being served better or worse over time.

VA increasingly relies on contracted medical providers to conduct disability exams, yet GAO found significant gaps in the agency's oversight of exam quality. Flawed medical exams produce flawed claims decisions, which can result in incorrect benefit determinations for veterans. GAO also identified separate gaps in how VA manages training for its claims processors. Both findings come from one of three prior GAO reports on which this testimony is based.

The AI Acceleration Problem

VA is exploring multiple uses of AI for disability benefits, including automating claims processing decisions. GAO has reported that generative AI can increase risk and hinder accountability, in part because even its designers may not fully understand how it works. The technology also requires significant computational and technical resources.

Applying AI to disability claims processing presents a specific challenge: detecting errors or misuse becomes harder because AI decision-making lacks transparency. GAO has developed an AI Accountability Framework to help agencies consider, select, and implement AI systems responsibly, and flagged it as a tool VA should use.

The Implementation Gap

Since 2021, GAO has made 43 recommendations to improve VA's disability compensation program. VA has implemented 28. The remaining 15 have been only partially addressed. GAO states that fully implementing them "would help VA better ensure veterans receive timely and appropriate benefits."

The 15 unresolved recommendations cover concrete operational problems: inadequate quality oversight, insufficient training, weak planning, and unclear performance metrics. An agency that has not closed out known, specific fixes is now moving to deploy AI at greater speed and scale.

The Bottom Line

VA must still fully implement the 15 recommendations that remain unresolved. Many trace back to GAO findings from prior years. GAO's AI Accountability Framework provides a structured path for responsible AI adoption, but its value depends on VA actually applying it, something the agency's track record on prior recommendations makes uncertain. For the 6.9 million veterans and their families depending on this system, the pace of that implementation is not an abstract management question.

Spot something wrong? Report an issue with this article