Written by: Aryan Samaga
Introduction
The first antitrust law, the Sherman Act, was passed in 1890 with the aim of protecting free and unfettered competition in the marketplace (“The Antitrust Laws,” 2025). The ambition was to put in place regulations that would prevent companies from colluding to artificially set prices and become overly dominant in their respective industries. In the years that followed, numerous pieces of legislation have been passed to uphold the integrity of competition in the free-market economy (Sawyer, 2019). But while we have seen multiple regulatory acts passed in the decades and centuries since that have built on top of the Sherman Act, namely the Clayton Act, the advent of AI has created numerous gray areas within legislation and raised the possibility that a raft of new legislation must be enacted to ensure the integrity of the free market remains intact.
Before the widespread adoption of self-learning algorithms (a subset of algorithms built using AI), algorithmic pricing legislation could be implemented and enforced with limited gray areas (Lardner et al, 2025). In the context of antitrust and competition law, a pricing algorithm is a computational process used by firms to determine prices for goods and services. To delve into the impacts of self-learning algorithms on competition and, ultimately, on competition law and regulation, this paper will first detail two distinct, albeit related, pricing algorithms and then articulate how self-learning algorithms have blurred the lines of legality for these algorithms. The paper will conclude with a discussion of the rhetoric surrounding potential policies to mitigate these concerns.
Monitoring Algorithms
To begin with, monitoring algorithms are broadly defined as algorithms used to collect, process, and use data (“Algorithms and Collusion: Competition Policy in the Digital Age,” 2017). In the pre-AI era, algorithms, even those with elements of what is now known as self-learning, were relatively easy to distinguish and hence to prove that companies were using them in a way that violated antitrust policy. Monitoring algorithms, specifically, are illegal when they are employed to facilitate the presence of an anticompetitive cartel (Boso Caretta & D’Andrea, 2023). Essentially, as long as monitoring is not used to uphold practices privy to those of a cartel, namely in monitoring the prices of other competitors to ensure that they are not breaking the bounds of a previous agreement, monitoring algorithms are legal. A typical example of a (legal) monitoring algorithm is when Company A notices that other retailers have reduced the prices of their TVs and, based on various data points and estimated supply and demand, decides to lower its own TV prices as well. Adjusting prices in this sense is perfectly legal, as Company A is adjusting prices not based on an agreement but on explicitly conducted market research.
Parallel Pricing Algorithms
On the autonomy scale, pricing algorithms can be viewed as a further extension of monitoring algorithms. Specifically, parallel pricing algorithms align a firm’s pricing schema with competitors’, leading multiple firms to move their prices together, even without direct contact (Boso Caretta & D’Andrea, 2023). The difference between a parallel pricing algorithm and a monitoring algorithm is that in a (legal) monitoring algorithm, a firm only keeps track of another company’s pricing for a good. However, in a parallel pricing algorithm, the company of interest not only monitors other companies prices for a said good but explicitly follows the price set by a separate company.
An example of this can be seen in the case of American and British online poster retailers making a pact to fix and maintain an algorithm to monitor competitors’ prices, essentially horizontally fixing prices (Nazzini & Henderson, 2022). Here, the retailers are explicitly using their algorithm to set and move their prices in tandem, while also working under a fixed agreement. This idea, moving prices solely in response to competitors’ pricing without any independent competitive justification, would constitute unlawful horizontal price-fixing, with the algorithm serving as the catalyst for the violation.
How Have Self Learning Algorithms and AI Blurred these Lines
In 2024, the FTC (Federal Trade Commission) announced its intention to introduce additional legislation targeting AI algorithms (Lardner et al., 2025). But why is that?
In a 2021 simulation, four independent Q-learning agents spontaneously learned collusive strategies in a pricing game (Calvano et al., 2021). The initial concern with the advent of AI in relation to pricing algorithms was that there would be more tools to circumvent antitrust laws and get away with violations. However, while that remains a concern, the larger issue is the possibility of “accidental collusion,” which is now perhaps the greatest challenge (Calvano et al, 2021). Namely, suppose these algorithms continue on their trajectory of improvement, and in turn, companies continue to look to them to best price their products. In that case, it is only a matter of time before the prices of goods all follow the same trends and are updated implicitly without any warning. This leaves the FTC and antitrust regulation in a bind because, unlike before, when a direct link between a collusive act could easily be identified, charged, and rectified, it is now much harder to determine whether collusion occurred by accident or on purpose. The current federal legislation lacks the capacity to address this, leading states such as New York to take early action before federal policy can catch up (Fitzgerald, 2025).
Potential Amendments to Policy
Traditional antitrust laws, such as the Sherman Act and its subsequent acts, were instituted to address human actions and communications, which, as a result, leave them incapable of addressing algorithmic collusions that occur without explicit coordination. This problem arises extensively with self-learning algorithms. One immediate solution would be for regulators to broaden the legal definition of “agreement” to require companies to audit and log their pricing systems. This would essentially put companies themselves at fault for any antitrust violations and could deter some companies from even trying to use these algorithms. Pushing these rules to third-party vendors could further limit the amount of collusion. Other ideas include ex-ante reviews and comprehensive reviews of high-risk algorithms, as well as a rebuttable presumption of collusion when dealing with any tool that consistently outputs supra-competitive prices (Mukherjee & Chang, 2025).
References
Boso Caretta, A., & D’Andrea, M. (2023). Algorithmic collusion. DLA Piper.
https://www.dlapiper.com/en/insights/publications/law-in-tech/algorithmic-collusion
Calvano, E., Calzolari, G., Denicolò, V., & Pastorello, S. (2020). Artificial intelligence, algorithmic pricing, and collusion. American Economic Review, 110(10), 3267–3297.
https://doi.org/10.1257/aer.20190623
Department of Justice, Antitrust Division. (n.d.). The antitrust laws and you. U.S. Department of Justice. https://www.justice.gov/atr/antitrust-laws-and-you
Fitzgerald, A. (2025, December 18). How companies can roll with New York’s new algorithmic pricing rules. Foley & Lardner LLP. https://www.foley.com/insights/publications/2025/12/how-companies-can-roll-with-nys-new-algorithmic-pricing-rules/
Lardner, A., et al. (2025, July 12). Antitrust and algorithmic pricing. The Regulatory Review.
https://www.theregreview.org/2025/07/12/seminar-antitrust-and-algorithmic-pricing/
Mukherjee, A., & Chang, H. (2025). Engineering rivalry: A “duty to disrupt” algorithmic convergence. SSRN.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5382575
Nazzini, R., & Henderson, J. (2024). Overcoming the current knowledge gap of algorithmic “collusion” and the role of computational antitrust (Vol. IV). Stanford Computational Antitrust, Stanford Law School.
https://law.stanford.edu/wp-content/uploads/2024/02/Algorithmic-Collusion.pdf
Organisation for Economic Co-operation and Development. (2017). Algorithms and collusion: Competition policy in the digital age. OECD Publishing.
https://www.oecd.org/competition/algorithms-and-collusion.html
Sawyer, L. P. (2019). U.S. antitrust law and policy in historical perspective (Harvard Business School Working Paper No. 19-110). Harvard Business School.
https://www.hbs.edu/ris/Publication%20Files/19-110_e21447ad-d98a-451f-8ef0-ba42209018e6.pdf
Youngson, N. (2013). Antitrust gavel. Independent Institute.
https://www.independent.org/article/2013/12/23/antitrust-busybodies/