HHS said on Thursday it is “supercharging” the way it reviews compliance audits for programs that receive federal health funding, expanding its use of artificial intelligence to help detect issues that officials say require attention. The department announced the new approach as it evaluates audit reports from states and other recipients on an ongoing basis, using tools that include ChatGPT to examine large volumes of documentation.
HHS’s finance leadership described the change as an effort to move beyond a system in which required audits can pile up without follow-through. Gustav Chiarello, the assistant secretary for financial resources who is leading the program, said the department plans to “dig into” audit submissions rather than treating them as paperwork that “lands with a thud and no one does anything about it.”
The initiative is designed around the federal audit requirement that applies to states, local governments, nonprofits and higher education institutions that spend at least $1 million in federal money each year. Those entities are required to submit annual audits, and HHS said the new initiative will use AI to analyze those audit reports from HHS-funded programs, including state Medicaid programs and federal grantees in areas such as research and addiction services.
HHS said recipients that do not file the required reports or do not resolve problems identified in them could face loss of funding. The department said it has already begun formal outreach, stating it sent letters to governors and treasurers in all 50 states to notify them of the new enforcement approach.
One of the letters reviewed by the Associated Press told state officials that HHS will no longer treat chronic audit noncompliance, repeat deficiencies, material weaknesses or delinquent audit obligations as matters that may remain unresolved through indefinite informal follow-up. HHS officials said the change reflects a broader push to pursue accountability in programs funded with federal health dollars.
The announcement also builds on HHS’s earlier adoption of generative AI for work tied to Medicaid investigations, administrative tasks and text editing, HHS said. In discussing the program’s safeguards, Chiarello said officials were evaluating public reports rather than uncovering new information, and he said the tools are intended to help grantees become better stewards of federal dollars.
The announcement comes as the Trump administration and Vice President JD Vance’s anti-fraud task force have promoted crackdowns in Medicaid and Medicare, as well as efforts aimed at student loan applications and other areas. The use of AI technology to flag likely fraud has been part of that broader effort, according to remarks by Federal Trade Commission Chairman Andrew Ferguson in a Fox News appearance.
Critics said the administration’s approach is not sufficiently concerned with fraud as such, and they questioned whether AI can be used fairly and nonpartisanly in the compliance process. Rob Weissman, co-president of Public Citizen, said he does not trust the administration to use AI tools in a fair and nonpartisan way, describing the AI as “kind of beside the point” in assessing what the administration’s objectives are.
HHS acknowledged to the Associated Press, in at least one instance, that it made a major mistake in data used to help justify a New York Medicaid fraud investigation. Asked about safeguards against the AI tools making mistakes, Chiarello said HHS was reviewing public reports and said the program’s intent was to support more effective compliance by recipients of federal funds.
At the same time, Chiarello said he has been in contact with counterparts in other federal departments, suggesting the department hopes other agencies would adopt similar technology. He said it would be “fairly easy” for other agencies to use HHS’s technology and “jump on it,” framing the move as a model the government could extend more widely.
In interviews and letters tied to the program, HHS presented the expansion as a targeted effort to increase enforcement against audit noncompliance and to identify issues earlier, using AI tools to process audit reports at scale. The department’s rollout, however, also underlines the challenge of using AI systems in government oversight—particularly when officials and critics raise concerns about accuracy, errors and potential biases.