The UK CMA is testing an AI tool to help catch companies that collude when bidding for public contracts, amid a “significant risk of bid-rigging” by contractors
Agency is trialling an AI-backed programme that scrapes data to identify bid-rigging Bluesky: @markcstanley . X: @procure4health Bluesky: Mark Stanley / @markcstanley : Hopefully they will continue to fine companies found to have participated. I guess the next question is how do companies who are bidding leverage AI to submit bids that avoid the appearance of a rigged bid! X: Rob Knott / @procure4health : Competition and Markets Authority is trialling a new artificial intelligence-backed tool it believes may help catch companies that collude when bidding for public contracts. “....real potential to drive billions of savings.” @HCSAprocurement https://www.ft.com/...
Context & Ripple Effects
The CMA has moved from its earlier fact-finding review of AI markets and scrutiny of major AI partnerships to a practical enforcement use case: analysing public-contract bidding. The trial matters because the authority identifies contractor bid-rigging as a significant procurement risk and sees detection as a route to lower public purchasing costs.
First-order effects
- The CMA can screen procurement data for patterns that may indicate coordinated bidding, prioritising suspect tenders for human investigation.
- Contractors bidding for public work face greater scrutiny of bid behaviour and the underlying records used to prepare and submit tenders.
Second-order effects
- Public buyers may need cleaner, more consistent tender data to make automated screening useful, increasing the value of procurement systems and audit trails.
- The prospect of data-led detection raises the cost of collusion: firms must strengthen competition-law controls around bidding, while legitimate bidders have more reason to document independent pricing decisions.
Third-order effects
- If the trial proves reliable, public procurement could become a major operational venue for state-mediated AI: not merely buying AI tools, but using them to monitor market conduct.
- The broader test will be whether automated signals can be made transparent and reviewable enough to support enforcement without treating pattern detection as proof of collusion.
The trend: Competition authorities are shifting from studying AI markets to applying AI within routine enforcement and public-market oversight.