Wonder why adoption of Artificial Intelligence technology is slow in public procurement? Most would tell you it’s the budget, or the bureaucracy, or the pace of government decision making holding up technology.
And yes, those are all very real problems facing public procurement across many issues. But there’s a bigger concern among professionals themselves, as well as their leadership, when it comes to adopting the most exciting new technology.
Of the 150+ procurement professionals surveyed in PlanetBids’ State of AI in Procurement survey, nearly half named accuracy and reliability of AI-generated outputs as a top concern. That number dwarfs every other response, including data privacy (40%) and legal or regulatory compliance risk (37%).
In open text responses, hallucination – AI generating confident but incorrect content – came up repeatedly. One respondent put it plainly:
In a private company, an AI error is a problem to be corrected. But in a public agency, it could mean a flawed solicitation, an incomprehensive evaluation, a vendor protest, or a public records request. The accountability standards that make public procurement trustworthy, like documented decisions, competitive processes, and fairness to all vendors, are exactly the standards that make AI errors highly consequential. Procurement professionals who have spent careers navigating these standards understand this intuitively. Their concern about accuracy is well founded, regardless of its potential for time savings.
The survey asked respondents whether they feel confident that they could defend an AI-assisted procurement decision in a vendor protest or public records request, and only 3% said they were very confident. 35% said they were concerned or very concerned, while another 35% hadn’t considered the question at all. Those 35% aren’t currently operating in an environment where AI use is widespread enough to raise the issue, but as AI continues to infiltrate the technology space and more agencies start adopting them, the question will come to everyone, whether or not they’re ready to answer it.
The good news is that the auditability gap isn’t a technology problem. It’s a process issue that’s solvable before you adopt any AI tool. Establishing a simple documentation standard for AI-assisted decisions that includes:
creates the audit trail you need to respond to a protest or records request. Build that process now, before you adopt a specific tool, so you’re not left designing a quick workflow under pressure while actively implementing new technology.
The full 2026 State of AI in Procurement Report covers the accuracy and trust findings from the survey in depth, including a practical audit trail template and guidance on what to require from vendors before adopting an AI-powered tool. Download the report now to see it all.