Overview
This is the second post in our series on antitrust and artificial intelligence. Our introductory piece highlighted four areas where AI intersects with competition law: agreements, unilateral conduct, mergers, and remedies. Here, we take up the first topic (AI agreements) and look at possible risks beyond the AI pricing tools that have dominated headlines and litigation to date. We look to RealPage and Gibson first as a baseline and then analyze risks in agreements beyond price fixing. While there is little precedent at this point, where AI facilitates anticompetitive agreements in areas such as market allocation, customer division, and information exchange, antitrust enforcers will step in to apply the antitrust laws.
I. Relevant Precedents: RealPage and Gibson
Recent cases underscore that algorithmic pricing is firmly on the radar of enforcers and courts. In United States v. RealPage, the Department of Justice alleged that multifamily landlords shared competitively sensitive, nonpublic data with RealPage's platform, which then generated rent recommendations for the multifamily landlords.1The DOJ argued this amounted to an unlawful agreement to align pricing even as it has conceded that landlords more often than not do not accept RealPage’s pricing recommendations for new leases.
The government’s theory is that each landlord that agrees to share its competitively sensitive data with RealPage "explicitly or implicitly understands and agrees that this arrangement involves other landlords likewise sharing their data with RealPage" and that those landlords "understand that they collectively benefit from this arrangement because it is unambiguously a mutual exchange of information among competitors."2Several defendants have entered into consent decrees with the DOJ in which, among other requirements, they agree not to use revenue management products that use non-public competitively sensitive data. The remaining parties are awaiting the court's rulings on their motions to dismiss.
Related enforcement efforts have included a parallel suit by the District of Columbia Attorney General and as well as a private class action in which the court granted a motion to dismiss plaintiffs' per se claims, but later denied a motion to dismiss plaintiffs' claims under the rule of reason standard.3These matters illustrate how both federal and state authorities as well as private plaintiffs are testing theories of AI-enabled coordination.
The Ninth Circuit’s August 2025 decision affirming the dismissal of the complaint in Gibson v. Cendyn Group illustrates the limits of the government's theory. In Gibson, plaintiffs challenged the use of Cendyn's hotel revenue management software, but the court dismissed the case because the AI system largely relied on public data and did not obligate hotels to adopt its recommendations. Without pooling of confidential information or mutual adherence to outputs, the court concluded there was no actionable "agreement" among horizontal competitors.4 The Ninth Circuit agreed, noting "plaintiffs push for a rule in which the choice of several competitors to contract with the same service-provider, when followed by higher prices, is sufficient to require antitrust scrutiny under the rule of reason," but concluding instead that "Section 1 requires a causal link between the contested agreement and an anticompetitive restraint of trade in the relevant market."[5]
Together, these cases sketch an emerging framework: it remains undecided, but likely, that AI may facilitate unlawful coordination when it aggregates confidential competitor data and pressures users toward common outcomes. But not every common use of AI, even if it produces parallel conduct, will lead to liability. For now, RealPage and Gibson dominate the landscape, and both remain focused on price. Thus, relevant precedent is limited, and because AI deployment across industries is still relatively new, the next generation of "agreement" cases could test novel theories regarding market division, customer allocation, and information exchanges.
The Ninth Circuit’s ruling in Gibson points to some factors to consider for how the antitrust laws may be applied to both pricing and non-pricing AI tools:
- Do you and your competitors use the same AI tool?
- Did you agree with your competitors to use a particular AI tool? If not, are there communications with your competitors about it?
- Did you communicate with the AI vendor about which of your competitors are using the AI tool (with the implication that the more participants that use it, the more powerful and useful it becomes)?
- Do you and your competitors feed the AI tool competitively sensitive information?
- Did you and your competitors agree to use the recommendations produced by the AI tool which are based on the competitively sensitive information fed by you and your competitors in a specific manner? If not, are there communications with your competitors about it?
II. Beyond Pricing: Other Agreement Risks
Although pricing tools have attracted the most attention so far, AI tools may also facilitate anticompetitive agreements related to the exchange of other competitively sensitive information. As with pricing information, when competitors feed confidential sensitive information into a shared AI model, the algorithm itself may function as a "hub," processing inputs and generating outputs that effectively reveal aspects of their rivals’ business strategies. Even if there is no specific agreement, the use of a common AI tool and the recommendations produced by the AI tool, could raise antitrust risks for those companies.
One possible risk involves AI tools that forecast customer demand or project market density based on competitively sensitive information. If competing firms all adopt the same system, they may independently decide to enter only underserved geographies and withdraw from areas where rivals are concentrated. This pattern could resemble a geographic market allocation scheme, even if no competitor ever explicitly agreed to stay out of another's territory.
Customer allocation is another risk. AI tools designed to optimize marketing campaigns may learn that head-to-head solicitation of the same customer segment reduces profits among competitors. Based on confidential inputs about the success of marketing efforts, a model could steer different firms toward distinct pools of customers after picking up patterns showing that non-overlapping marketing efforts are the most successful. If competitors broadly adopt such a tool, the result could be functionally identical to a non-solicitation agreement, reached not in the haze of a smoke-filled room, but through reliance on a shared algorithm that recommends taking action for potentially unlawful reasons.
Yet another risk presents itself when companies use AI to limit production of goods and the provision of services in ways that keep supplies low and prices high. The riskiest and most foreseeable scenario may be trading platforms that aggregate companies' confidential sales and order data. If existing personal finance and stock trading platforms are any guide, there will be a strong market incentive for these platforms to use the wealth of data in their systems to add value for their users by recommending trading strategies and transactions. If a critical mass of a market’s participants use the same trading platform, it is easy to see how such features could encourage competitors to adopt strategies that maximize their individual and collective profits by, for example, cutting production or timing trades, in ways that would clearly constitute anticompetitive collusion if done by any other means. Commodity transactions and other areas with high sales volumes could conceivably fall into such a pattern with a common AI platform.
At least under the decisions issued to date, these scenarios would be limited to situations in which AI models generate output to competitors based on confidential, competitively sensitive information supplied by competitors. Merely engaging in parallel conduct based on the ways in which AI may revolutionize the ability to process and make use of publicly available data would not necessarily be problematic in the current landscape. However, there are other areas outside of information-sharing among competitors that could still raise risks when using AI, including when firms with market power use AI in an anticompetitive way to further increase an already high market share. The use of an AI projection alone likely would not be illegal, but, for example, if a firm with market power used AI-generated market density projections and then took some anticompetitive actions to block its competitors based on the AI recommendations, it could give rise to concerns.
Also, there are risks in the way that AI can reshape vertical agreements. A manufacturer that requires its distributors to adopt its proprietary AI platform for managing sales may, in effect, hardwire resale terms into the AI tool. If the system embeds preferred pricing or steers distributors toward particular sales channels, the result could be a new variant of resale price maintenance (which remains a per se violation of the antitrust laws of several states). Even without an explicit contractual restriction, the AI system itself may limit the independence of downstream firms.
The DOJ, FTC, the UK Competition and Markets Authority, and the European Commission, have all highlighted their concerns for risks arising from agreements in the generative AI space. In a joint statement in July 2024, enforcers on both sides of the Atlantic warned that collusion among key players could stymie challengers and "steer market outcomes." That statement noted that partnerships, financial investments, and other connections between firms related to the development of generative AI "have been widespread to date" and "could be used by major firms to undermine or coopt competitive threats and steer market outcomes in their favor at the expense of the public."6
III. Conclusion
AI is reshaping how companies interact with competitors, customers, and distributors, and antitrust law is only beginning to grapple with its implications. As of today, there are only a handful of cases, and little precedent on how courts will treat AI-facilitated agreements in other contexts.
This is a new frontier, but the lesson from RealPage and Gibson is clear: when AI substitutes for the “meeting of the minds,” courts and enforcers will not hesitate to apply antitrust laws. These agreements are apt to extend well beyond price fixing, encompassing market allocation, customer division, and information exchange. A useful rule of thumb for antitrust and AI was coined by former FTC acting Chair and Commissioner Maureen Ohlhausen: "[i]f it isn't OK for a guy named Bob to do it, then it probably isn't OK for an algorithm to do it either."7
In our next post, we will turn to unilateral conduct and examine how AI use by a dominant firm could be viewed as exclusionary.
1United States et al v. RealPage, Inc., No. 1:24-cv-00710, Amended Complaint, ECF 47 (M.D.N.C. Jan. 7, 2024).
2United States et al v. RealPage, Inc., No. 1:24-cv-00710, Plaintiff’ Response in Opposition to Defendants MTDs at 4, ECF 145 (M.D.N.C. May. 29, 2025).
3District of Columbia v. RealPage, Inc., No. 2023-CAB-6762, Complaint, (D.C. Super. Ct. Nov. 1, 2023); In Re: Real Page, Inc., Rental Software Antitrust Litig (No. II), 709 F. Supp. 3d 478, Order Denying Motion to Dismiss (M.D. Tenn. Dec. 28, 2023).
4 Gibson v. Cendyn Group, LLC, No. 2:23-CV-00140-MMD-DJA, 2024 U.S. Dist. LEXIS 83547 (D. Nev. 2024).
5Gibson, et al. v. Cendyn Group, LLC, et al., No. 24-3576, Opinion at 6, 2025 WL 2371948 (9th Cir. Aug. 15, 2025).
6Joint Statement on Competition in Generative AI Foundation Models and AI Products (Jul. 2024), available at: https://www.gov.uk/government/publications/joint-statement-on-competition-in-generative-ai-foundation-models-and-ai-products/joint-statement-on-competition-in-generative-ai-foundation-models-and-ai-products.
7Maureen Ohlhausen, Acting Chairman, Fed. Trade Comm’n, Remarks at the Concurrences Antitrust in the Financial Center Conference: Should We Fear the Things That Go Beep in the Night? Some Initial Thoughts on the Intersection of Antitrust Law and Algorithmic Pricing (May 23, 2017) at 10 available at https://www.ftc.gov/system/files/documents/public_statements/1220893/ohlhausen_-_concurrences_5-23-17.pdf.