A late January deadline for federal agencies to submit a list of AI use cases passed without the Small Business Administration meeting it. In March, the SBA entered the fray by releasing its inventory for the first time in its history.
How the agency responded to the release of the 2025 inventory is emblematic of how the agency has approached federal AI compliance generally, according to a new report from the Government Accountability Office, which chided the SBA for “not consistently following” reporting requirements for the use of emerging technologies.
Although the SBA finally published a list of AI use cases two months ago, the requirements for agencies to publicly report their use have been “in the works since 2020,” the report said. The U.S. Advancement in AI Act of 2022 required government agencies to publish their inventories annually.
Nevertheless, GAO says it took until this spring for the SBA to comply with the law, despite the fact that it has been using a form of this technology since at least 2021, when it introduced machine learning tools aimed at preventing fraud in its pandemic loan programs.
That use case has since been discontinued, according to the report, but guidance from the Office of Management and Budget states that “agencies must report not only current use cases, but also those that are being piloted or not yet fully implemented, as well as those that have been discontinued or discontinued within the past year.”
SBA shared its 2024 internal inventory with GAO listing 21 use cases. Officials told the watchdog that all of these uses “will either be phased out, not implemented, or suspended as of March 2025.”
The pause was intended to give the agency time to review its use cases’ compliance with the new executive order and reflect “shifted priorities” by the SBA’s new chief information officer and chief AI officer. The agency’s CIO position was flipped several times during the early years of the second Trump administration, before being replaced by Hartley Caldwell, a former fintech and IT executive who also serves as the SBA’s CAIO.
Last month, SBA officials told GAO that while they were still updating AI-related policies to comply with the EO, the agency’s Data Governance Committee and AI Governance Council were working on pilot projects to “test the security, value, and performance of various AI capabilities.”
“Examples include pilot programs designed to test AI’s ability to report fraud using specific SBA program datasets, as well as various pilot or pre-pilot use cases related to IT operations and cybersecurity launched in June, September, and December 2025,” the report states.
The watchdog recommended that SBA’s CIO establish policies to meet federal AI reporting requirements and document implementation of these steps and other significant decisions. The report said authorities agreed with this recommendation.
“Establishing policies and procedures for reporting AI use cases, including defining roles and responsibilities and documenting their implementation, will ensure that SBA reports applicable AI use inventory,” the report states. “Until the SBA does so, transparency will remain limited and public and congressional oversight of information sharing and agency use of AI will be hampered.”
According to recently released OMB data, the use of AI across the federal government will increase by 70% from 2024 to 2025, with approximately 3,600 use cases reported across agencies last year.
