When Algorithms Set Prices: AI and Antitrust Law in Conflict
By Dirk Röthig | CEO, VERDANTIS Impact Capital | March 7, 2026
November 2025: The U.S. Department of Justice settles with RealPage. The software company had provided thousands of landlords with an AI-powered pricing system — the result: coordinated rent increases across millions of American apartments, without the landlords ever speaking to each other. No back-room meeting, no secret phone call. Just an algorithm. This case is no American anomaly. It is the blueprint for one of the most pressing legal questions in the digital economy — and it is landing in Europe with full force.
Tags: Competition Law, Antitrust Law, Artificial Intelligence, Algorithms, EU Law
A Landlord Who Never Spoke to His Competitors
At the heart of the RealPage scandal lies a seemingly simple business model: A software company collects price data from competing landlords, feeds it into an algorithm system, and returns individual price recommendations to each landlord. The recommendations were based on the prices of direct competitors — without the landlords themselves ever making contact with each other.
The U.S. Department of Justice (DOJ) saw this as illegal price-fixing under antitrust law. In its lawsuit from August 2024, joined by attorneys general from seven states, the agency argued: If competitors use the same AI pricing platform and supply it with non-public price information, this creates de facto coordinated behavior — even without an explicit agreement (DOJ, 2024). In November 2025, the parties agreed to settle with far-reaching obligations: monitoring by an independent monitor, complete disclosure of records to authorities, cooperation in ongoing proceedings against other landlords (Wilson Sonsini, 2025).
The message to industry was unmistakable: An algorithm does not protect against antitrust liability.
The Legal Core: What Is Actually Prohibited?
To understand the scope of this development, one must examine the European legal framework. Article 101 of the Treaty on the Functioning of the European Union (TFEU) prohibits all agreements, decisions, and concerted practices between undertakings that appreciably restrict competition. The decisive term is "concerted practice."
The Court of Justice of the European Union (CJEU) has established in its case law: Such concerted practice does not require a formal agreement. It suffices that undertakings know or can reasonably be expected to know that their market partners will exhibit the same behavior — and they act accordingly (CJEU, 1999, C-199/92 P — Hüls AG). This definition, originally developed for classic cartel cases, applies to algorithmic price coordination with striking precision.
The European Commission confirmed this assessment in the revised Horizontal Guidelines of June 1, 2023. The Guidelines — the central interpretation document for horizontal competition agreements — identify three critical scenarios of algorithmic price coordination that may fall under Article 101 TFEU (European Commission, 2023).
Three Paths to Antitrust Liability
Scenario 1: The Direct Algorithm Cartel
The clearest scenario is also the one most similar to classic cartels: Competitors explicitly agree to use the same algorithm for price-setting. If rivals in food retail, gas stations, or hospitality engage the same price optimization provider and supply it with identical cost data, this can be deemed illegal information exchange under competition law — a violation "by object," sanctionable without proof of concrete market effects (European Commission, 2023).
Scenario 2: Hub-and-Spoke Coordination
More complex — and practically more common — is the hub-and-spoke model. Here, the algorithm provider acts as the "hub," while the using companies form the "spokes." Each spoke provides the hub with sensitive competitive data, the hub processes it, and returns optimized prices to each participant. No spoke speaks directly to another — but all benefit from the shared information pool.
The European Commission and several national cartel authorities agree: This model too can fall under Article 101 TFEU, provided the participants know or should know that their competitor uses the same platform (Morgan Lewis, 2025). The RealPage case was a textbook example of this construct.
Scenario 3: Tacit Collusion Through Machine Learning
The most technically demanding — and for regulators the most difficult to grasp — is so-called "tacit collusion by algorithm." Here, algorithms are not programmed for collusion. They develop collusive pricing strategies through independent learning.
Laboratory experiments and theoretical models show: When two self-learning pricing algorithms compete in a market with repeated interactions, they converge without any human intervention on supra-competitive prices (Calvano et al., 2020). They "discover" independently that mutually maintaining prices is more profitable than price undercutting — similar to how human oligopolists behave in certain market structures.
Whether this behavior falls under Article 101 TFEU remains legally unresolved. The German Federal Cartel Office acknowledges the problem: In its discussion paper "Algorithms and Competition," the authority states that collusion can demonstrably arise under laboratory conditions — whether this is transferable to real markets remains an open question (Federal Cartel Office, 2021). The Expert Circle on AI and Competition, which the office convened in June 2025, put this question back on the agenda without reaching a conclusive result (Federal Cartel Office, 2025).
The Burden of Proof Question: How Does One Prove Algorithmic Collusion?
Here lies the fundamental dilemma for cartel authorities. Classic cartels leave traces: emails, meetings, internal memos. Algorithmic coordination, by contrast, can arise completely without documentation. No executive need ever have spoken with a competitor. No algorithm developer need have intended collusive behavior. The outcome — coordinated prices to the detriment of consumers — emerges as an emergent phenomenon from market dynamics and machine learning.
For cartel authorities, this means: They must either prove that the companies deployed the algorithm with knowledge of its coordinating effect, or they must develop new liability models not based on subjective fault. The European Commission is experimenting with the concept of "Compliance by Design": Companies should be required to demonstrate that their algorithms have no collusive tendency — a burden of proof reversal that is controversial in legal scholarship (Hogan Lovells, 2025).
The British Competition and Markets Authority (CMA) took a more pragmatic approach in 2024: It launched the "Dynamic Pricing Project," a cross-sector investigation of algorithmic pricing from e-commerce to energy supply, and announced "Do's and Don'ts of Dynamic Pricing" for 2025 — behavioral rules that should serve companies as safe harbor guidance (CMA, 2024).
German Competition Law: GWB Meets AI
In Germany, the Act Against Restraints on Competition (GWB) applies, which serves as the national equivalent to Articles 101/102 TFEU. The 10th GWB Amendment of 2021 already substantially digitalized German antitrust law: Market power can now be assessed appropriately for platform markets and data ecosystems. But for algorithmic price agreements — particularly tacit collusion through self-learning algorithms — there is still no clear legal framework.
The Federal Cartel Office has signaled that it intends to pursue algorithmic coordination using existing instruments, even if the legal classification is complex. President Andreas Mundt has emphasized in various interviews that the office applies behavioral rules for humans to the deployment of algorithms: If a human would act illegally under antitrust law, this remains illegal when an algorithm does the same (LTO, 2023).
What Companies Must Do Now
The regulatory pressure on algorithmic pricing systems is no longer a future scenario — it is present reality. According to its own statements, the European Commission is conducting multiple sector investigations into algorithmic pricing starting in July 2025 (Global Competition Review, 2025). This creates concrete compliance obligations for companies:
1. Algorithm Audit: Every AI-powered pricing system should undergo antitrust legal review. Central questions: What competitor data flows in? Can the system generate price recommendations that effectively replicate competitor prices?
2. Data Governance: Sensitive competitive data (prices, capacities, margins) should not flow into shared algorithm platforms without antitrust review. The mere fact that a third party processes the data does not absolve companies of their liability.
3. Documentation of Independence: Companies should document that their pricing decisions are made independently of competitors — even if they use external AI systems.
4. Regulatory Monitoring: The legal situation is evolving rapidly. The CMA's "Do's and Don'ts," the new EU Guidelines, and national GWB developments require continuous monitoring by antitrust-specialized advisors.
Outlook: The End of Regulatory Blind Spots
Dirk Röthig, CEO of VERDANTIS Impact Capital, summarizes the situation as follows: Digitalization has exponentially increased the speed of business decisions — the law always lags somewhat behind. On the topic of algorithmic pricing, this gap is closing very quickly. Companies that today still believe their AI systems are antitrust-neutral could discover otherwise tomorrow.
The regulatory development shows a clear direction: Cartel authorities in the USA, EU, and UK no longer view algorithmic pricing as a technical edge case, but as a central enforcement field. The RealPage settlement with its far-reaching obligations — independent monitoring, data protection rules for competitor information, agency cooperation requirements — is likely to serve as a blueprint for future European proceedings.
For practitioners, this means: AI systems are not legal vacuums. Those who let algorithms set prices bear full and immediate antitrust responsibility for doing so. The technical complexity of the systems does not protect against liability; it merely increases the burden of proof for authorities. In the long run, this is not sustainable protection.
More Articles by Dirk Röthig
- AI-First Companies: How Native AI Firms Disrupt Industries — How AI-native companies structurally challenge traditional industries
- Evaluating AI Investments: A Framework for VC and PE — Five dimensions and ten critical questions for AI investments
- Computer Vision in Industry: Quality, Logistics, Safety — Machine vision in manufacturing practice
References
Federal Cartel Office (2021): Algorithms and Competition. Digital Series, No. 6. Federal Cartel Office, Bonn. Available at: www.bundeskartellamt.de
Federal Cartel Office (2025): Expert Circle on AI and Competition — Press Release 24.06.2025. Federal Cartel Office, Bonn. Available at: www.bundeskartellamt.de
Calvano, E., Calzolari, G., Denicolò, V., Pastorello, S. (2020): Artificial Intelligence, Algorithmic Pricing, and Collusion. American Economic Review, 110(10), pp. 3267–3297.
CMA — Competition and Markets Authority (2024): Dynamic Pricing — Inquiry Announcement. London: CMA.
DOJ — U.S. Department of Justice (2024): Justice Department Sues RealPage for Algorithmic Pricing Scheme that Harms Millions of American Renters. Washington D.C.: DOJ Office of Public Affairs, August 2024. Available at: www.justice.gov
CJEU (1999): Hüls AG v. Commission, C-199/92 P. Court of Justice of the European Union, July 8, 1999.
European Commission (2023): Guidelines on the application of Article 101 of the Treaty on the Functioning of the European Union to agreements on horizontal cooperation between competitors. Official Journal of the EU, June 1, 2023. Available at: eur-lex.europa.eu
Global Competition Review (2025): EU competition authorities zero in on antitrust risks of algorithmic pricing. London: GCR, July 2025. Available at: globalcompetitionreview.com
Hogan Lovells (2025): Update on algorithmic pricing in competition law — What you need to know. Washington D.C./London: Hogan Lovells. Available at: www.hoganlovells.com
LTO — Legal Tribune Online (2023): Antitrust Law: When the Algorithm Sets Prices. Berlin: LTO. Available at: www.lto.de
Michigan Journal of Economics (2026): History of Pricing Algorithms & How the Newest Iteration has Antitrust Policy Scrapping for Answers. Ann Arbor: University of Michigan, January 2026. Available at: sites.lsa.umich.edu
Morgan Lewis (2025): Algorithmic Pricing Emerges as Enforcement Priority for EU & UK Antitrust Regulators. Philadelphia/London: Morgan Lewis & Bockius, October 2025. Available at: www.morganlewis.com
Wilson Sonsini Goodrich & Rosati (2025): DOJ Settles Its Algorithmic Price-Fixing Case Against RealPage. Palo Alto: Wilson Sonsini, November 2025. Available at: www.wsgr.com
About the Author: Dirk Röthig is CEO of VERDANTIS Impact Capital based in Zug, Switzerland. As an entrepreneur and investor, he engages with the legal and economic frameworks of the AI economy, sustainable impact investments, and the intersection of technology, regulation, and competition law. Further articles and contact: www.verdantiscapital.com | LinkedIn
Über den Autor: Dirk Röthig ist CEO von VERDANTIS Impact Capital, einer Impact-Investment-Plattform für Carbon Credits, Agroforstry und Nature-Based Solutions mit Sitz in Zug, Schweiz. Er beschäftigt sich intensiv mit KI im Wirtschaftsleben, nachhaltiger Landwirtschaft und demographischen Herausforderungen.
Kontakt und weitere Artikel: verdantiscapital.com | LinkedIn
Top comments (0)