SBA Flagged for Noncompliant AI Fraud Detection Pilot
The SBA deployed Palantir tools to flag borrower fraud without meeting federal AI risk management standards.
Updated on Sept. 29, 2026 in Artificial Intelligence

Live Poll
Should federal agencies face stricter oversight when using AI to manage public benefits and programs?
The Small Business Administration utilized Palantir software to identify fraud in pandemic-era loans without properly documenting the pilot as a high-impact use case. This failure to meet Office of Management and Budget requirements led to the suspension of 111,620 California borrowers.
Why it matters
The case highlights the growing tension between rapid AI adoption in federal agencies and the necessity of adhering to established governance frameworks to prevent automated administrative errors. As federal bodies integrate complex models into benefits systems, the failure to identify high-impact tools threatens the legal and financial security of thousands of citizens.
The agency-run pilot utilized Palantir software to surface data and leads for criminal enforcement, yet it failed to meet mandatory risk mitigation standards set by the Office of Management and Budget. While 111,620 loans were suspended in California, the lack of an AI governance board means these actions currently operate without a formal impact assessment.
The players
Small Business Administration
A federal agency that provides support to entrepreneurs and small businesses and administers disaster and pandemic-era loan assistance programs.
Palantir
A data analytics company that provides software platforms for integrating and managing massive datasets to support large-scale institutional decision-making.
Office of Inspector General
An independent oversight body responsible for auditing federal agency programs and ensuring compliance with regulatory and legal requirements.
The details
The Small Business Administration used software from Palantir—a developer of data analytics and decision-support platforms—to process loan information and flag potential fraud for criminal investigators. However, the agency failed to categorize the initiative as a high-impact use case, a classification requiring stricter transparency and documentation under federal guidelines. Without this oversight, the program moved to formal status in July 2026 without conducting the impact assessments necessary to safeguard against errors in automated loan suspensions.
Timeline
April 2025: OMB released a memo requiring AI risk mitigation steps.
January 2026: SBA signed a contract with Palantir for fraud detection.
February 2026: SBA suspended 111,620 California borrowers for suspected fraud.
April 2026: OIG completed its review of the SBA fraud program.
July 2026: SBA formalized its use of the Palantir AI software.
The Tech Race
This development follows a pattern where federal agencies adopt advanced AI tools before formalizing internal governance structures required by the April 2025 OMB memo. The ongoing conflict between agency operational speed and compliance mandates remains the primary hurdle for institutional AI deployment.
The 111,620 suspended borrowers in California now face significant uncertainty due to the absence of consistent remedies for those caught in the flawed detection process. Future applicants for federal loans should monitor for changes in agency transparency as the Small Business Administration addresses the OIG's recommendations.
The takeaway
The Small Business Administration must now reconcile its operational speed with mandatory risk mitigation procedures to satisfy OIG oversight. Observers should track whether the agency successfully establishes an AI governance board to address the five pending recommendations.
What happens next
The House of Representatives has passed legislation requiring the Small Business Administration to submit a formal report on its AI usage to Congress.
Further reading
For more on federal AI oversight, read the latest analysis from our Artificial Intelligence section.
Source note: This article includes information reported by Nextgov.
Live Poll
Should federal agencies face stricter oversight when using AI to manage public benefits and programs?









