AI and public procurement in Spain: from anti-collusion screening to secure automation

The gradual incorporation of artificial intelligence into public procurement is a reality that is transforming the way administrations and companies operate. The debate no longer centres on whether  artificial intelligence (AI) tools should be used, but on how they should be integrated and where their limits should lie. Recent examples in Spain and other EU Member States demonstrate that the integration of AI is reshaping public procurement.

Overview of the use of AI in public administrations

Spanish administrations have significantly accelerated the integration of advanced analytics technologies into their procurement systems. This trend is not exclusive to Spain, as the Organisation for Economic Co-operation and Development (OECD) has also documented similar initiatives in countries such as Ukraine, Chile, Brazil and Portugal.

In Spain, the most ambitious initiative has come at the national level. The government has announced plans to transform the public procurement platform by incorporating artificial intelligence, automation, and big data analytics. The aim is to detect irregular patterns and anticipate risky behaviour.

This strategy aligns with recommendations from the European Commission, the Group of States Against Corruption (GRECO) and the OECD, particularly given recent mismanagement episodes that have reignited social concerns about the transparency of public processes.

The regional landscape is also changing rapidly. For instance, Asturias has adopted the Tendios platform, an AI-powered digital tool designed to optimise public procurement for government entities and private companies. The aim is to automate the preparation of supporting documents and specifications, streamline processing, and improve the traceability of files. Another example is the Generalitat Valenciana, which has chosen AI systems that are specifically designed for fraud prevention and the early detection of irregularities, thereby reinforcing a procurement model that is geared towards integrity.

These initiatives confirm that, while AI is being deployed asymmetrically according to organisational capabilities, it is inevitably advancing towards all phases of the contract cycle.

For example, before tendering, these types of AI tools can be used to facilitate market analysis, define requirements and create more consistent and comparable specifications. During the competitive phase, they streamline the classification of bids, the detection of anomalous values and the review of documentation. In the execution phase, they can be used to evaluate milestones, identify deviations, monitor subcontracting patterns and segment risks in real time.

AI applied to public procurement

In the context of concerns over recent episodes of corruption in procurement, AI appears to be a tool capable of rebuilding trust. Its ability to analyse thousands of files, prices and contractors simultaneously and detect atypical patterns could transform oversight in public procurement. Public administrations would be able to identify suspicious rotations, documentary coincidences, circular subcontracting and price variations that previously went unnoticed.

In this regard, Spanish competition authority, the CNMC, has taken a step forward with BRAVA (Bid Rigging Algorithm for Vigilance in Antitrust), an antitrust surveillance algorithm designed to detect collusive practices in public tenders. Its ability to classify bids opens up a field that, until now, depended almost exclusively on manual evidence or investigations. Previously, these techniques focused on numerical variables, but generative AI now enables the incorporation of elements that were previously unmanageable.

For years, many anti-competitive behaviours went unnoticed, such as the construction cartel in Spain producing virtually identical technical reports. Today, generative AI makes it possible to compare large volumes of documentation, identify stylistic similarities and replicated fragments, and evaluate textual patterns. What would once have required an almost heroic manual review can now be detected in seconds.

However, AI is not just a tool for regulators. In the private sector, particularly among small and medium enterprises, it is beginning to be used to assist with tendering processes. Historically, the complexity of tender documents and the associated bureaucratic burden have excluded operators with fewer resources from participating in public procurement.

Against this backdrop, various tools have emerged that use AI specifically in the public procurement sector to democratise the market. These tools enable real-time tracking of tenders, summarise specifications using natural language models, and generate personalised alerts. As a result, companies that were previously unable to participate can now do so with confidence.

The central question about the scope of AI

Public procurement is an area in which strict rules and technical assessment criteria coexist. While AI promises to reduce uncertainty and make decisions more objective, it can also eliminate the room for consideration that characterises administrative discretion. This tension partly stems from the fact that algorithmic systems reduce the discretion of bodies that previously assessed facts or evaluated evidence according to their own criteria. Now, algorithms can be proposed as mechanisms for detecting patterns based on large volumes of data, which can create the impression of greater neutrality.

Yet this objectivity has nuances. AI can only work if someone decides which objectives to optimise or which data is relevant. In public procurement, for example, there will always be an element of human judgement. Hence, the idea of a "humanity reserve" is beginning to take hold: technology can assist, but it cannot make decisions in areas where the law requires assessment and responsibility on the part of the contracting body.

In the EU, the AI Act classifies public procurement as a high-risk use, requiring transparency, traceability and ongoing oversight. In addition, standard contract clauses for the purchase of AI systems were published in March 2025. These clauses introduce requirements for explainability, auditing, human control and risk management: thereby avoiding situations in which the administration acquires tools that it cannot understand or control (a situation that would pose a real risk and which must be avoided).

The challenges of implementation

Using AI tools alongside digitalisation multiplies the ability to detect conflicts of interest, cartels, anomalous behaviour and systemic irregularities. However, it also introduces challenges that should not be overlooked and requires robust implementation guidance. Dependence on technology providers, potential biases, inadequate explanations and blind trust in automated indicators can create new forms of inequality, opacity and breach of safeguards.

Incorporating AI into a highly regulated environment subject to the principle of equality poses a legal and ethical challenge due to the introduction of automation. The principal challenge lies in ensuring the explainability and auditability of algorithms to prevent bias or unfair exclusion, particularly when they are employed for ranking purposes.

While the European framework establishes supervisory bases, Member States and autonomous administrations must develop their own rules and protocols to complement algorithmic governance, as well as training for the professionals who will be responsible for using these types of tools.

Osborne Clarke comment

Despite the uncertainties surrounding the implementation of AI tools in public procurement, we believe that this trend is irreversible and cannot be ignored. It is becoming increasingly clear that AI is rapidly transforming public procurement. While this transformation has positive aspects, such as promoting integrity, democratising access and modernising processing systems, it also requires new skills, new rules and a firm commitment to transparency and reasoning. In future, the key will be to balance innovation with guarantees and legal certainty.

If you would like to know more about the issue discussed in this note or other issues related to public procurement, please do not hesitate to contact one of our experts listed below or your usual contact at Osborne Clarke.