Companies using algorithmic pricing tools should carefully consider antitrust implications. There are several ways to mitigate or reduce the antitrust risk. Typical ways to limit exposure include implementing internal controls to monitor and/or document any unintended price alignment resulting from the algorithm.
Companies increasingly deploy artificial intelligence (AI) and algorithmic pricing software tools to optimize their sales. While these tools can deliver substantial efficiencies and advance competition, they also carry the potential to be used in ways that restrict competition. Sometimes technology can be two steps ahead of internal compliance measures.
Antitrust framework
The cartel prohibition provision of Article 101 of the Treaty on the Functioning of the European Union (TFEU) prohibits and penalizes all agreements and concerted practices between undertakings (meaning: economic operators of any kind) that have an effect on trade between EU Member States and whose object or effect is to distort competition. The national equivalent in the Netherlands of this provision is Article 6 of the Dutch Competition Act (DCA).
An illegal agreement or practice does not have to be between competitors. An intermediary can also be held liable under law when facilitating prohibited conduct. However, an anti-competitive agreement can be exempted if it provides benefits to those impacted by a loss of competition and meets other criteria (e.g., joint production agreements between rivals leading to innovations for the benefit of customers).
For reference, dominant firms (simplified: those companies with a market share of approx. 40% or more) are bound by even stricter rules under Article 102 TFEU and/or Article 24 DCA.
European Commission
Based on public statements of the European Commission (EC), the EC appears to be highly suspicious of algorithmic pricing tools. The EC’s main concern in relation to pricing algorithms is their potential ability to facilitate coordinated conduct, resulting in higher prices, including by means of collusive agreements. According to the EC, even the use of only publicly available data may be an antitrust violation in certain instances. For example, if an algorithm (using only public data) is being used by two firms to ‘peg’ prices with respect to one another, the EC may infer the existence of agreements between those firms if the ‘cost’ at which the information is accessed is reduced by the algorithm. The EC is pushing the envelope of current antitrust case law. If the same algorithm is used by several firms, the EC may allege that common use of algorithmic software is an agreement to establish a starting price, which would violate the antitrust laws. Agreeing to jointly use algorithmic pricing software might also be targeted as a restriction of competition. The EC is currently investigating matters involving algorithmic pricing tools.
Dutch Competition Authority
In July 2025, the Dutch Competition Authority (ACM) announced that it intends to conduct a sector inquiry into computer-controlled consumer prices in the aviation sector in the Netherlands. A sector inquiry typically assesses whether a market is functioning well for consumers and identifies any bottlenecks. Once the investigation has been completed, the ACM can make recommendations to the legislator to minimize or resolve any potential problems. If the ACM has evidence of antitrust violations, it can also take enforcement action itself. Any insights regarding pricing algorithms could potentially also be used in other sectors at a later stage.
Conclusion
Algorithmic pricing tools are under increased antitrust scrutiny by the EC and the ACM. Companies using or considering the use of those tools are advised to conduct careful antitrust analysis. There are several ways to mitigate or reduce the antitrust risk. Typical ways to limit exposure include ensuring that the algorithm does not have access to non-public data and implementing internal controls to monitor and/or document any unintended price alignment resulting from the algorithm. Any evidence that the algorithm creates price volatility or leads to an overall price reduction could also be helpful.
