Why public procurement may decide the future of AI accountability

The research introduces the framework of transformative law, a concept that views law not merely as a mechanism that regulates technology, but as a strategic instrument capable of shaping innovation and institutional evolution. AI procurement, in this sense, becomes a fulcrum through which States influence how emerging technologies are designed, deployed, and governed.

Governments around the world are accelerating digital transformation, but new research shows that the way public institutions purchase artificial intelligence may determine far more than procurement efficiency. A growing body of legal analysis now suggests that AI procurement is becoming one of the most powerful mechanisms for reshaping how the modern State functions, regulates, and governs.

A new scholarly article, “Transformative Public Procurement of Artificial Intelligence,” published in Laws, argues that public contracts for AI systems are no longer routine administrative tools. Instead, they are emerging as strategic legal infrastructures capable of guiding technological innovation, steering institutional reform, and shaping the trajectory of public governance itself. 

AI procurement: A legal instrument capable of structural change

The author explains that artificial intelligence is no longer a peripheral tool but a system-transforming force reshaping analytical processes, decision structures, and operational models within governments. As AI becomes embedded in public administration, the legal system must confront fundamental questions about accountability, transparency, democratic oversight, and the protection of rights.

The research introduces the framework of transformative law, a concept that views law not merely as a mechanism that regulates technology, but as a strategic instrument capable of shaping innovation and institutional evolution. AI procurement, in this sense, becomes a fulcrum through which States influence how emerging technologies are designed, deployed, and governed.

The author argues that public procurement should no longer be regarded as a transactional procedure aimed solely at purchasing goods or services. Instead, it must be understood as a structural tool that enables public authorities to direct technological development toward democratic values and rule-of-law principles. This interpretation positions procurement as a form of regulatory leverage, allowing States to embed substantive legal safeguards into the architecture of AI systems before they enter public service.

This conceptual shift rests on a fundamental premise: if the State must comply with principles such as algorithmic transparency, human oversight, and anti-discrimination when using AI, then those principles must be built into the procurement process itself. Procurement, therefore, becomes the legal gateway through which the normative foundations of the State are preserved in an era of digital transformation.

AI's dual role: A tool for procurement and an object of transformation

The study identifies two distinct but interconnected roles that AI plays in public procurement.

First, AI can be used within procurement procedures. Governments increasingly deploy algorithmic systems to streamline tender evaluations, detect irregularities, reduce corruption risks, and manage large datasets. When AI functions as an internal procedural tool, it influences the integrity and efficiency of procurement processes themselves.

Second, AI is also the object of procurement, with public authorities purchasing systems to support or automate public functions. This second role carries far greater structural implications. Public-sector AI may affect decisions in social services, justice, public safety, education, employment regulation, and critical infrastructure. Each deployment alters how public functions are conceptualized and executed.

These two roles raise complex regulatory challenges because they require the State to ensure algorithmic legality both in its internal processes and in the systems it acquires. This includes the need to guarantee non-exclusive machine decision-making, adequate human oversight, safeguards against discriminatory outcomes, and mechanisms ensuring interpretability.

The study finds that public procurement is increasingly viewed as a means not simply to acquire AI technology but to shape its qualities. Contractual clauses, tender specifications, performance monitoring mechanisms, and post-award obligations collectively influence the design and behavior of AI systems. In this sense, procurement becomes a formative stage in determining how AI operates when performing public functions.

Challenges, asymmetries, and regulatory complexity in AI procurement

While the study highlights procurement’s transformative potential, it also identifies deep structural challenges in implementing this model.

One major issue is the asymmetry of expertise between public agencies and private suppliers. AI developers possess superior technical knowledge, which can allow them to influence contractual terms, technical standards, and performance expectations. This imbalance risks allowing private interests to shape public governance outcomes, undermining democratic oversight.

Even when contractual clauses attempt to impose transparency or oversight requirements, the practical ability to enforce them depends heavily on the public authority’s organisational capacity. The author stresses that contract performance, not merely contract award, becomes a decisive stage, since many AI system qualities only become measurable during implementation. Data governance, monitoring, validation, and internal expertise become essential components.

Another challenge arises from the regulatory framework introduced by the EU AI Act, which establishes additional obligations when public authorities procure or deploy “high-risk” AI systems. The Act imposes requirements related to risk management, data quality, robustness, accuracy, transparency, and human oversight. These obligations apply to both suppliers and public bodies, creating a dual accountability structure.

Because classification as “high-risk” depends on specific definitions and exceptions, the regulatory landscape becomes complex. The study notes that interpretive uncertainties may create inconsistent practices, especially for public purchasers who must determine whether a system falls within the high-risk categories. As a result, procurement processes are now embedded within a multilayered legal environment that spans data protection, cybersecurity, administrative law, and sector-specific regulations.

The research warns that broad, principle-based contractual templates, including updated EU model clauses for AI procurement, may not fully protect public institutions from market-driven standards set by technologically dominant suppliers. Without stronger administrative regulation and institutional capacity, procurement risks generating fragmented or insufficient governance outcomes.

AI procurement as legal infrastructure for future governance

The study proposes reinterpreting AI procurement as infrastructure, not merely a sequence of contracts but a systemic architecture shaping how public functions evolve over time.

Based upon social science theories of infrastructure, the author argues that procurement embeds AI systems into the core machinery of the State. These systems influence decision-making processes, administrative behavior, public-private interactions, and the organisation of public functions. Once deployed, they require continuous data curation, regulatory oversight, technical maintenance, and institutional coordination.

Under this infrastructural model, procurement is not a one-time activity. It becomes a long-term governance framework involving multiple actors, public bodies, suppliers, regulators, auditors, and civil society. Each contract contributes to a broader ecosystem that defines how AI shapes public authority. The transformation occurs not only through immediate technological capabilities but also through the cumulative institutional pathways the procurement system establishes.

To sum up, the transformative potential of AI procurement lies in its dual role as both a tool for acquiring technology and an instrument for designing the future operational structure of the State. If executed strategically, procurement can foster accountability, ensure compliance with democratic values, and promote responsible innovation. However, without adequate regulatory coordination and institutional capacity, AI procurement may instead amplify risks, entrench imbalances, or create opaque systems that limit democratic control.