How Public Sector Procurement Should Think about AI

Technological innovation is no longer confined to specific industries or market niches. And public administrations can no longer be mere spectators. They are, in fact, central players in shaping how these technologies are deployed, governed and leveraged. 

Traditionally, the public sector has exercised two primary functions in relation to innovation.  

First, as a regulator. This role is particularly critical in the case of AI, where the need for ethical safeguards, human-centric design and risk mitigation is crucial — a priority now crystallized in the EU’s pioneering AI Act, the world’s first attempt to regulate AI comprehensively. 

And second, as an accelerator. Through funding mechanisms, incentive schemes and targeted support programs, governments have the capacity to stimulate adoption and scale innovation beyond what the private sector could achieve alone. By strategically allocating resources, the public sector helps ensure that technological benefits are broadly distributed rather than concentrated in the hands of a few.  

In recent years, a third role has emerged with increasing significance: tech user.  

Tech Use in the European Public Sector 

More and more public administrations are adopting and integrating advanced technologies into their daily operations. From digital public services to data-driven decision-making, governments are experimenting with tools that can enhance efficiency, transparency and citizen engagement.  

These three roles – regulator, accelerator and user – define a more complex relationship between the public sector and technology. Governments are actively shaping the trajectory of technological change. 

In this way, procurement is pivotal. Indeed, public procurement represents one of the most powerful tools available to governments: it is no longer a purely administrative exercise, but rather a policy instrument. By adopting AI responsibly, for example, governments can model best practices and send strong market signals, encouraging private actors to follow similar standards of safety, accountability and transparency. AI can fundamentally reshape how administrations purchase goods and services, monitor compliance, detect inefficiencies and manage increasingly complex supply chains.   

Europe stands at the forefront of defining how technology can serve the public good. Through initiatives such as the EU AI Act, the Digital Europe Programme and the European Data Strategy, the EU has laid the groundwork for a distinctive approach to innovation — one that combines competitiveness with accountability, and progress with protection of fundamental rights.  

Six Ways AI Can Be Applied in Public Sector Procurement 

Public administrations across Europe are now translating these principles into practice, demonstrating that digital transformation can advance not only efficiency but also trust, fairness and inclusion. Several European countries, including Finland and Estonia, are leading the way in applying AI to public procurement. 

The following are ways public administrations can boost their procurement functions with AI: 

1. Analysis and evaluation. Public contracts generate an enormous volume of data, and AI can help administrations detect meaningful patterns, anticipate future demand and verify the accuracy of the information in procurement platforms. It can also monitor tender documents to ensure greater clarity and quality of language. Some governments are already moving in this direction. In South Korea, a pilot project has applied deep learning to forecast the government’s annual demand for products, drawing on more than ten million records from its product management system. In Finland, Hansel Oy – the central purchasing body of the Government, developed a platform to analyze prices more effectively, offering a database of historical costs and advanced applications for cost engineering. 

2. Categorization of public spending. By automating the categorization of public spending, governments can prevent irregularities, reduce the risk of fraud and corruption and ensure that procurement operates in a transparent and reliable manner. Accuracy rates of AI-driven classification systems are estimated to be at 97%. In Ukraine, the ProZorro Platform uses machine learning to automatically predict the correct Common Procurement Vocabulary Code based on tender information, saving time for procurement officials, broadening supplier participation and fostering fairer competition. In Australia, a central purchasing body developed the AI tool Caity to classify spending data according to the public procurement taxonomy. What once required the manual processing of over two million transactions every six months is now completed automatically, freeing resources for higher-value tasks.  

3. Fight against corruption. By processing vast quantities of procurement data and detecting complex patterns, AI can help authorities focus their attention on the cases that truly matter. This shift enables governments to move from reactive investigations to proactive prevention of corruption, using technology to shine a light on risks that would otherwise remain hidden. The stakes are high: according to the International Monetary Fund, corruption costs governments worldwide between $1.5 and 2 trillion every year. Several countries have already begun to deploy AI in this domain. In Brazil, tools such as ALICE, which automatically flags anomalies in bids, MONICA, which provides visibility on procurement data (such as suppliers contacted most frequently) and SOFIA, which analyzes reports to identify errors and suspicious correlations, have been rolled out to strengthen oversight. In Ukraine, the DoZorro Platform has demonstrated impressive results in identifying high-risk tenders: during its beta-test, it was able to detect 26% more cases of incorrect winner selection, 37% more unjustified disqualifications, and nearly 300% more instances of collusion among bidders. These examples show how A.I. can be used not only to improve efficiency, but also to safeguard integrity in public procurement.   

4. Drafting procurement documents. By drawing on benchmarks and previous experience, AI tools can support the preparation of standardized templates and ensure that requirements are formulated in the clearest and simplest possible terms. This not only reduces the administrative burden on contracting authorities but also makes procedures more accessible and understandable for economic operators, thereby fostering broader participation and fairer competition. In the U.S., the Department of Health and Human Services has launched the HHS Accelerate project that provides procurement officials with tailored recommendations on the standard clauses to include in the documents. This approach streamlines the process, ensures greater coherence, and allows human officials to focus on the strategic dimensions of procurement rather than on administrative tasks. 

5. Chatbots. By handling frequent questions, these tools free up staff time and allow employees to focus on higher-value activities. At the simplest level, FAQ bots can understand common queries and deliver pre-set answers. Virtual assistants go further, not only responding to questions but also carrying out basic actions. More advanced still are virtual agents”, capable of engaging in complex dialogue and effectively replacing human staff in specific functions. A notable example is Procurement Answers and Information Guided Experience, which was developed by the City of San Francisco to respond to about 1,000 procurement-related questions, drawing on a knowledge base of some 400 predefined answers.  

6. Automating tedious processes. Image and video analysis, for example, offers powerful opportunities for supplier oversight. Algorithms can detect defects, assess risks and identify areas for improving product quality – tasks that would otherwise require considerable human effort. Concrete steps are already being taken in this direction. In Finland, the State Treasury is preparing to use AI to automate invoice processing. The system can scan vast amounts of data in seconds, match invoices with contract numbers and amounts and then channel them directly into electronic payment flows.  

The Risks of AI in Public Sector 

It's important to remember that the same technologies that make procurement more efficient, transparent and user-friendly also introduce new vulnerabilities. As public administrations become active users of AI, they must also confront the risks it brings. Among the issues most frequently cited are intrusive surveillance, erosion of privacy, potential job losses, the spread of fake news and disinformation, increased discrimination, and even the distortion of policy-making process.  

The irony, of course, is that while governments rush to deploy AI in procurement, they are also racing to regulate it. The AI Act is a reminder that public administrations are both experimenters and watchdogs — a dual role that risks becoming schizophrenic if not handled with care. Legislation and regulation remain essential tools to prevent misuse, for example, by strengthening privacy laws and making certain harmful applications of AI illegal. Yet, given the speed of technological evolution, legislation alone is not sufficient.  

Safety measures must also be reinforced by subjecting technologies to stricter testing, monitoring and oversight. And, while public agencies should adopt principles and guidelines that can steer AI development and deployment in ways consistent with public values, these can be difficult to implement, as they often lack the hard incentives that would encourage adherence.