knowledge 21 May 2026 |

Levelling the playing field? The growing use of AI by competition authorities

Anti-competitive practices are adapting to exploit emerging technologies including artificial intelligence. At the same time, competition authorities in the EU and beyond are themselves utilising AI to gather market intelligence, increase efficiency and improve antitrust enforcement in an ever-changing digital landscape. The use of pre-emptive monitoring mechanisms paired with the ability of authorities to analyse vast amounts of information faster than ever before will likely result in greater market oversight with a consequent increase in new investigations and dawn raids. We explore these developments below.

The Michelin Decision

The European Commission’s investigation into tyre manufacturers represents the most high-profile instance of AI-assisted antitrust enforcement. The European Commission suspected that several companies in the tyre industry, including Michelin, may have used public communications to signal and therefore coordinate future sales prices (in particular, wholesale prices) for replacement tyres for cars and trucks in the EEA. The suspected practices included the deliberate use of public communications to inform undertakings about future pricing intentions and strategies so as to influence pricing policies. The Commission used AI to assist in analysing hundreds of thousands of public statements and earnings calls (followed by a qualitative review) to flag potential signals of coordination, which led to a dawn raid being carried out on the premises of Michelin (and several other tyre companies) in January 2024.

Following the raid, Michelin challenged the decision before the General Court, claiming, amongst other matters that the Commission lacked sufficiently serious indicia to justify the inspection.

The decision of the General Court implicitly validated the use of AI to screen public information as the basis to carry out a dawn raid, finding that the approach adopted by the Commission sufficient to justify the inspection for the main period. Crucially, what is clear from the judgment is that it was the combination of quantitative analysis (carried out by AI) together with manual qualitative analysis that rendered the screening methodology legitimate.

While the General Court emphasised that there must be “sufficiently serious indica” suggesting an infringement that specifically corresponds to the time period under scrutiny, the case illustrates that the Commission now has the tools to analyse extensive amounts of publicly available data. As a result, it is likely that we will see a more proactive approach to competition law enforcement and much more cases in the future.

The General Court partially upheld Michelin’s appeal. It annulled the inspection decision concerning the earlier period of suspected infringement on the basis that the Commission’s screening did not uncover “sufficiently serious evidence to support suspicions of price coordination during” that period, but confirmed the legality of the inspection for the main period. 

AI Practices of Certain Competition Authorities

National competition authorities are increasingly looking to incorporate AI into their work processes.

One of the authorities at the forefront of these developments is the Spanish National Markets and Competition Authority (CNMC), which has deployed an AI-assisted tool known as BRAVA (Bid Rigging Algorithm for Vigilance in Antitrust). Trained on the Spanish public procurement database, BRAVA relies on machine learning to analyse and classify bids to assess whether they are potentially collusive. In February 2026, Cani Fernández, CNMC President, stated BRAVA has already been used in cases resulting in sanctions, and that the CNMC is liaising with the US Department of Justice to provide training on the technology. The initial success of the project has also led to CNMC tendering its first AI contract for the further development of AI instruments.

Similarly, the Lithuanian Competition Council has piloted the use of AI-driven e-discovery tools to assist in cartel detection, including an AI-enhanced document review platform designed to identify signs of collusion. Through contextual and behaviour analysis, the platform can identify unusual communication patterns. Trial runs showed that the model could complete an evidence retrieval task in one week, with the typical timeline for similar exercises being several months. The overall results of testing were described as mixed, but promising.

Outside the European Union, the UK Competition and Markets Authority (CMA) is trialling a new AI tool to identify bid-rigging and collusion in bids for public contracts. Public procurement accounts for nearly a third of all public spending in the UK, and this project aims to reduce expenditure and increase productivity in the CMA. The pilot programme was described as successful.

The Brazilian Administrative Council for Economic Defence (CADE) is seeing clear results from the Cerebro project, commenced in 2014 to research and develop technological solutions for the investigation of cartels. Databases were formed through web scraping (an automated process of extracting data from websites using software or scripts) and partnerships with national and local authorities. This data was systemically organised into firm-centric categories such as corporate structures and IP addresses, and tender-centric categories such as bidding categories and reoccurring co-bidders. AI machine-learning tools were then utilised to analyse procurement documents for any signs of collusion.¹ The detection of unusual patterns among companies operating in the highway engineering sector resulted in Operation Novo Rumo, which eventually led to proceedings being brought against sixteen companies and fifteen individuals.

Although technical details remain largely confidential, trends can be seen in the use of document and data analysis to detect patterns pointing towards bid-rigging or other collusive behaviours.

The Digital Transformation in Competition Law Enforcement Project

An understanding that not all national competition authorities are equally prepared to adopt AI resulted in the launch of the Digital Transformation in Competition Law Enforcement (DICE) project in January 2026. Co-funded by the EU under the Technical Support Instrument,² this scheme aims to reduce gaps in digital enforcement capacity and enhance the AI capabilities of 15 competition authorities across the EU. The project will design and deliver practice-oriented training including Massive Open Online Courses (MOOCs), workshops, hybrid training sessions and hands-on labs.

Ensuring all competition authorities are well-equipped to tackle the challenges posed by AI will prevent regulatory fragmentation and guarantee the consistent application of competition laws across the EU.

Conclusion

The incorporation of AI into regular enforcement practice is a large undertaking, but competition authorities across the globe are recognising that it is becoming increasingly necessary. While authorities are by no means equally advanced in the development and deployment of these new technologies, a general willingness to adapt and embrace new operational methods can clearly be seen across the board.

Also contributed to by Amy Gebruers