Image: ocp-website-production-2fo5ied6.imgix.net · rights & removal

AI won’t fix corruption in public procurement, but it may help you catch it

Reporting by Open Contracting PartnershipRead the original at open-contracting.org

Executive Summary

Public procurement represents a major risk area due to opaque, paper-based systems lacking effective oversight for trillions of dollars spent by governments. Global norms are shifting toward greater transparency, focusing on quality over sole price selection and single bidding, supported by data availability through standards like the Open Contracting Data Standard (OCDS). Artificial intelligence is entering this space by enabling the analysis of messy, unstructured documents to detect risks. Various AI applications currently include using Large Language Models for drafting documents, Retrieval-Augmented Generation to check bids against regulations, machine learning for fraud detection, and exploratory uses of AI agents. Practical applications involve extracting linked data across documents (e.g., asset disclosures and campaign donations), flagging procurement anomalies based on corruption indicators, profiling supplier risks using financial data, and detecting bid-rigging patterns at scale through advanced pattern recognition. While AI can enhance detection, it requires investment in structured data to measure market performance effectively. The ultimate goal involves rethinking procurement processes and reducing bureaucratic friction, embedding AI into systems like the DREAM ecosystem for real-time project management while maintaining human accountability.

Facts Only

* Trillions of dollars are spent through opaque, siloed, paper-based public procurement systems lacking effective oversight.
* The 2023 UN resolution promoted transparency and integrity in public procurement, including technology guidelines for combating corruption.
* Procurement agencies use Large Language Models (LLMs) to generate text from static knowledge, such as drafting Request for Proposal (RFP) templates or extracting data.
* Retrieval-augmented generation is used to check bids against national regulations based on defined legislative corpora.
* Machine learning and algorithms compare patterns and data for fraud detection.
* AI agents are explored to facilitate processes like auto-uploading documents or emailing winners.
* Use cases include extracting, matching, and linking data from structured and unstructured documents, such as asset disclosures and campaign donations.
* Systems have been developed to flag procurement anomalies based on corruption risk indicators in Chile, Brazil, and Kazakhstan.
* Supplier risks are profiled by integrating beneficial ownership data with public registries.
* Machine learning detects bid-rigging patterns by analyzing massive datasets for price analysis and bidder relationships.

Full Take

The narrative frames AI as a powerful tool to bridge the gap between regulatory intent (transparency) and systemic reality (corrupt practice), moving from reactive detection to proactive prevention, particularly in high-volume processes like procurement. The tension lies in the necessary integration of opaque algorithmic decision-making with fundamental principles of accountability. The piece rightly identifies that AI's utility stems from its ability to find patterns in existing data rather than creating new truths; this necessitates a commitment to structured data standards like OCDS, which is where the initial measurement framework must be built. A critical pattern emerges around the inherent risks of automation: if systems automate high-stakes decisions (like flagging anomalies or processing contracts), the safeguards—audit trails, explainability, and liability—must be designed into the architecture from inception, rather than being retrofitted. The emphasis on human oversight is a necessary friction against potential systemic failure arising from black-box reasoning or poor data quality. The underlying assumption that technology alone is a panacea overlooks the prerequisite shift in institutional friction and public will required for successful governance transformation. What are the specific mechanisms by which accountability is guaranteed when decision-making shifts to machine logic? How can the risk of outputting unreliable results—due to poor input data or model error—be systematically managed without sacrificing operational efficiency?

From the original · Open Contracting Partnership

There is a reason public procurement is the government’s biggest corruption risk. Trillions of dollars are spent by governments through opaque, siloed and paper-based systems with no lever for effective oversight.

Read the full story at open-contracting.org

Sentinel — provisional

Raw score, 0 to 1 (uncalibrated)0.15

No strong signs of machine writing were found in the source article. Provisional estimate, not a finding that a person wrote it.

The model's own note (unverified)

The article presents a nuanced argument about AI in public procurement, effectively balancing potential benefits with critical implementation risks, supported by specific international examples.

This looks only at the wording of the original source article, not at this page's AI-written sections. A small local AI model made this estimate. It has not been checked against known human and machine texts, so treat it as provisional. It cannot show who wrote an article.

What the model noticed in the source article

Sentence length variance shows natural variation; sophisticated vocabulary is used effectively.

The text maintains a coherent argument flow, moving logically from problem to application to limitations.

Specific examples (Kazakhstan, Chile, Brazil) are integrated coherently with the theoretical discussion, suggesting human synthesis rather than template matching.

Claims about specific projects and regulatory frameworks appear grounded in cited examples, showing careful contextual integration rather than pure fabrication.

Signs the model read as against machine writing

The text incorporates direct references to specific international bodies (UN resolution, World Bank) and country-specific initiatives (ChileCompra, ALICE), indicating grounded research.

The transition between theoretical discussion and concrete case studies is handled with a reflective, persuasive tone characteristic of expert commentary.