SBA Can't Account for Its Own AI Use Even Though It's Used to Oversee Billions in Contracts: GAO

Why it Matters

The federal government spends hundreds of billions of dollars each year on contracts, and small businesses are supposed to get a meaningful slice of that. Now, artificial intelligence is being positioned as a tool that could reshape how those contracts are awarded, how fraud is caught, and how innovation programs are run. But a new federal watchdog report finds that the agency most responsible for championing small businesses in that system can't even account for how it is using AI itself.

The Government Accountability Office released a report on May 4 examining how AI could be deployed across AI small business contracting operations, federal innovation research programs, and the Small Business Administration's own offices. The findings carry real consequences: without transparency about how AI is being used inside federal agencies, small businesses, Congress, and the public have no way to assess whether the technology is helping or causing harm.

Broader Context

The SBA sits at the center of a sprawling federal effort to direct contracts and research funding toward small businesses. Every major federal agency is required to maintain an Office of Small and Disadvantaged Business Utilization, known as an OSDBU, coordinated by SBA. These offices are the internal advocates ensuring small businesses are considered when agencies spend money.

Separately, the Small Business Innovation Research and Small Business Technology Transfer programs, known as SBIR and STTR, funnel billions of dollars in federal research funding to small firms developing cutting-edge technologies. Federal agencies with extramural research and development budgets exceeding $100 million are required by law to allocate 3.2 percent to SBIR and 0.45 percent to STTR.

The GAO report examined how AI could be applied across all three of these areas. The potential applications identified are broad: conducting market research, reviewing proposals, analyzing data, drafting mandated reports, and flagging potential fraud. The case for AI fraud prevention contracting is particularly relevant given the scale of federal spending flowing through these programs.

The Risks

The GAO did not simply catalog AI's upside. The report identified three categories of risk that federal agencies and SBA must contend with as they consider AI federal contracting applications.

The first is inaccurate outputs. AI systems can generate plausible-sounding but factually wrong information, a phenomenon sometimes called hallucination. In a contracting context, an AI system that misreads a proposal or misidentifies a vendor's eligibility could steer awards in the wrong direction.

The second is data security. Federal contracting involves sensitive information about vendors, pricing, and procurement strategies. Feeding that data into AI systems raises questions about how it is stored, who can access it, and whether it could be exposed or misused.

The third is bias. AI systems trained on historical data can replicate and amplify existing patterns of discrimination. In small business contracting, where disadvantaged businesses already face structural barriers, biased AI outputs could quietly undercut the very equity goals these programs are designed to advance.

SBA's Transparency Problem

The report's most pointed finding concerns SBA's own conduct. Under the Advancing American AI Act, enacted in December 2022, federal agencies are required to publicly publish inventories of their AI use cases. The law was designed to give Congress and the public visibility into how the government is deploying artificial intelligence.

SBA has not been meeting that obligation. The GAO found that the agency lacks policies and procedures for publicly publishing its AI use case inventory. The report states directly that "without policies and procedures for publicly publishing its AI use case inventory, SBA may continue to risk not complying with the Advancing American AI Act," and that this failure "hinders transparency and congressional and public oversight."

The GAO recommended that SBA develop and implement those policies and procedures. The finding is notable because SBA is the agency charged with coordinating AI small business contracting oversight across the federal government, yet it cannot demonstrate accountability for its own AI activity.

Congressional Response

Lawmakers have been pressing on AI governance across multiple fronts, and the SBA's compliance gap has attracted direct legislative attention.

H.R. 8664, introduced in the 119th Congress, would require the SBA Administrator to implement a specific recommendation related to artificial intelligence adoption, language that tracks closely with the GAO's findings. The bill reflects congressional frustration that voluntary compliance with transparency requirements has not been sufficient.

In February 2026, leaders of the House Science, Space, and Technology Committee separately asked GAO to review the broader federal and state AI regulatory landscape, signaling that oversight of federal agency AI implementation is becoming a sustained congressional priority, not a one-time inquiry.

Strained Programs

The report lands at a moment when the SBIR and STTR programs are already under strain. The programs experienced a roughly five-month lapse in early 2026, with approximately $4 billion in funding frozen during that period, according to reporting by Granted AI. That disruption left small businesses and research institutions in limbo, and it has sharpened scrutiny of how these programs are managed.

Against that backdrop, the question of how AI is being used, or misused, inside federal innovation research programs carries added weight. If agencies are deploying AI to review proposals or allocate resources without adequate oversight or transparency, the risks identified by GAO, including bias and inaccurate outputs, could compound the damage already done by the funding freeze.

The Bottom Line

The GAO's core argument is straightforward: AI is already being considered for use across federal small business programs, and the public deserves to know the details. The Advancing American AI Act created a mechanism for that transparency. SBA's failure to use it is not a technical oversight. It is a governance failure with practical consequences for the small businesses, researchers, and entrepreneurs who depend on these programs.

The existence of H.R. 8664 suggests that at least some members of Congress are not willing to wait for SBA to act on its own.

Access the Legis1 platform for comprehensive political news, data, and insights.