Pricing Algorithms and Antitrust: Regulating AI-Driven Coordination

Price algorithms are computerized formulas that determine a price automatically based on a set of data inputs. [1] They have been used in a variety of industries for decades; however, technological advancements have created a shift from mechanically set prices to AI-powered algorithms. [2] These AI algorithms can handle large amounts of data and have the capability to interact, learn, and make quick decisions. [3] As price algorithms become more advanced, antitrust regulators have become more concerned, stating they facilitate explicit and tacit collusion. [4] While explicit algorithmic collusion can generally be prosecuted under current antitrust law, algorithmic tacit collusion presents a challenge, suggesting that regulation of firms’ algorithms may be necessary to protect competition and consumers.  

Algorithmic explicit collusion occurs when multiple business entities have explicitly agreed upon the level of their prices and further agree to use price algorithms to monitor and adjust their prices, or when multiple firms use price algorithms to coordinate pricing with each other. [5] AI pricing algorithms facilitating explicit collusion can primarily be seen in United States v. David Topkins. [6] David Topkins, a director of a company selling posters online, was held liable for horizontal price fixing with other merchants on Amazon’s platform, where Topkins and his co-conspirators agreed to fix prices of certain posters sold on Amazon in the United States. [7] To implement this agreement, they adopted specific price algorithms for the agreed-upon posters, with the goal of coordinating changes to their respective prices. [8] Topkins wrote computer code that instructed the algorithm-based software to set prices of the agreed-upon posters in conformity with this agreement. [9] Ultimately, the Court found Topkins and his co-conspirators in violation of Section 1 of the Sherman Act which prohibits contracts, combinations, or conspiracies that restrain trade or commerce among the states or with foreign nations. [10] However, Topkins took a plea deal agreeing to pay a $20,000 criminal fine, to face 6 to 12 months of imprisonment, and to cooperate with the U.S. Department of Justice’s investigation. [11] Despite the relatively low penalties, United States v. Topkins is considered a landmark legal case because it was the first time that the Division prosecuted defendants where AI was a tool to further antitrust misconduct. [12]

While explicit collusion can easily be proven through direct communication between firms, tacit collusion involves coordination without direct communication making it much harder to prosecute. [13] Unlike traditional explicit collusion that may generate meeting minutes, messages, or other documentary traces, algorithmic interactions often leave limited traces or no record. [14] Therefore, if tacit collusion were to happen through interaction between AI self-learning algorithms, both parties may not even know tacit collusion is occurring. [15]

These risks become evident through different scenarios in which algorithmic tacit collusion might occur. The first scenario where tacit collusion may occur is the ‘Hub and Spoke’ model, where multiple competitive firms use a common price algorithm to coordinate prices among themselves. [16] The most extreme scenario of Hub and Spoke occurs when firms entrust their pricing policy to the same provider of algorithmic pricing services. [17] This results in coordination by the third-party agent through the collection of data and the application of the price algorithm without the firms’ knowledge. [18] The second scenario is simpler, with firms using simple pricing algorithms reacting to market conditions in a predictable manner. [19]  For example, firms using the tactic of ‘lowest price matching’—a competitive strategy where firms commit to matching any lower price set by competitors—can lead to transparent and predictable pricing, resulting in tacit collusion and parallel pricing. [20] The third scenario involves self-learning technology, where algorithms are complex enough to learn by themselves. [21] When firms program these algorithms with the objective of profit maximization, the price algorithm may use trial and error to determine the most efficient pricing to align with its competitors to maximize profits. [22] These scenarios demonstrate the ways in which algorithmic tacit collusion may occur; however, the ability of existing antitrust law to address such conduct has been inconsistent.

The Sherman Act outlaws every contract, combination, or conspiracy in restraint of trade, and any monopolization, attempted monopolization, or conspiracy or combination to monopolize. [23] The Clayton Act addresses specific practices that the Sherman Act does not clearly prohibit, such as mergers and interlocking directorates. [24] The Federal Trade Commission (FTC) has been active in evolving the interpretation of tacit collusion under U.S. law. [25] In the late 1970s to 1980s, the FTC attempted to challenge tacit collusion under Section 5 of the FTC Act, prohibiting unfair methods of competition in violation of the Sherman Antitrust Act or Clayton Act. [26] However, after several unsuccessful cases against the oil, fuel additives, and cereal industries, the FTC abandoned using Section 5 to challenge tacit collusion, stating that it “reaches beyond the Sherman and Clayton Acts to encompass various types of unfair conduct that tend to negatively affect competitive conditions”. [27] Consequently, the lack of enforcement of the Sherman and Clayton Acts will promote the continuation of algorithmic tactical collusion; therefore, antitrust laws must be updated for the AI digital economy, otherwise, more consumers may continue to see higher prices. [28] While explicit collusion can be addressed under antitrust law, tacit collusion is likely beyond the reach of them as currently enforced. [29] However, a potential solution for governments in addressing advancing price algorithm technology is regulation. [30]

According to MacKay and Weinstein in “Dynamic Pricing Algorithms, Consumer Harm, and Regulatory Response,” there are two possible ways to regulate firms’ price algorithms. One potential regulation of price algorithms is by controlling when firms can set prices. [31] For example, firms would be required to price only once a day or once a week at the same time. [32] Though regulating pricing frequency would not address commitment to reacting to rivals’ price changes, it could eliminate firms’ ability to employ strategies that may appear competitive but generate higher prices, drawing a clearer line between competitive and collusive conduct. [33] Another option is to implement a rule where firms are prohibited from incorporating rivals’ prices into their algorithms. [34] This would disrupt a leader-follower pattern as superior algorithms would be prevented from automatically undercutting prices set by inferior algorithms. [35] Furthermore, it would allow firms to re-price as often as they seem fit, letting them quickly react to market conditions, excluding rivals’ price changes. [36] While this method might reduce firms’ ability to compete on price, firms’ algorithms would have an acceptable amount of data to work with, therefore, not inhibiting firms’ algorithms severely. [37]

Regulation efforts have already begun with California legislation enacting Assembly Bill 325 (AB 325) and Senate Bill 763 (SB 763). [38] AB 325 adds Section 16729 to the Business and Professions Code, broadly prohibiting the use or distribution of “common price algorithms” in anticompetitive environments. [39] AB 325 defines a common pricing algorithm as “any methodology, including a computer, software, or other technology, used by two or more persons, that uses competitor data to recommend, align, stabilize, set, or otherwise influence a price or commercial term.”. [40] Furthermore, it also bans the use of shared pricing algorithms containing competitor data even if it’s publicly available. [41] AB 325 also adds Section 16756.1 to the Business and Professions Code, establishing a new plea standard for Cartwright Act claims. [42] Under this new standard, a complaint will be sufficient to survive dismissal on the pleadings if it alleges facts that make a conspiracy plausible, protecting plaintiffs from early dismissals. [43] SB 763 significantly increases criminal penalties for antitrust violations from $1 million to $6 million for corporations and up to $1 million per violation for individuals, deterring firms from violating antitrust law. [44] Additionally, the bill also imposes new civil penalties of up to $1 million per violation in cases brought by the California attorney general or a district attorney. [45] The implications of AB325 can be illustrated through the case Machi v. Yardi Systems Inc. In the case, the California state court granted summary judgment to the defendants, concluding that the software’s functionality did not breach state antitrust and unfair competition law, as it did not implement non-public competitor data to suggest prices. [46] However, AB 325’s prohibition of the use of common price algorithms as well as publicly available competitor data would alter this analysis, potentially exposing similar pricing software to greater antitrust scrutiny.

The widespread use of price algorithms in today's market has major implications for consumers, businesses, and policymakers. These price algorithms provide more efficiency for businesses, but also create new challenges for antitrust enforcement. Without any regulation on price algorithms, the risk of collusion and, therefore, consumer harm will continue. However, antitrust regulators have already started to address challenges that algorithmic tacit collusion brings through legislation such as California's Assembly Bill 325. Regulation ensures that antitrust laws keep pace with advancing technology, protecting competitive environments and consumers.

---

Sources

1. Alexander MacKay and Samuel Weinstein, “Dynamic Pricing Algorithms, Consumer Harm, and Regulatory Response,” Washington University Law Review 100, no. 1 (2022): 113  https://doi.org/10.2139/ssrn.3979147. 

2. Stephanie Assad, Robert Clark, Daniel Ershov, and Lei Xu, “Algorithmic Pricing and Competition: Empirical Evidence from the German Retail Gasoline Market,” Journal of Political Economy 132, no. 3 (2024): 723 https://www.journals.uchicago.edu/doi/10.1086/726906

3. Ibid

4. MacKay and Weinstein, “Dynamic Pricing Algorithms,” 114.

5. “Antitrust Implications of Using Pricing Algorithms,” David Cardwell and Tom Pickard, Chambers and Partners, July 19, 2023,https://chambers.com/articles/antitrust-implications-of-using-pricing-algorithms.

6. Ibid

7. United States v. Topkins, No. 15-cr-00201, at 3 (N.D. Cal. 2015).

8. Ibid.

9. Ibid.

10. Ibid. 

11. Ibid.

12. “AI and Antitrust – When Does an Algorithm Become an Agreement?,” Jennifer Driscoll, JD Supra, May 19, 2023,https://www.jdsupra.com/legalnews/ai-and-antitrust-when-does-an-algorithm-6819337/.

13. “Antitrust Implications of Using Pricing Algorithms,” Antonina Yaholnyk and Anastasia Zeleniuk, Chambers and Partners, March 13, 2020,https://chambers.com/articles/antitrust-implications-of-using-pricing-algorithms.

14. “Algorithmic Pricing and Antitrust Risk,” Ross Ferguson, Benjamin Gris, Annie Herdman, Nicole Kar, Henrik Morch, Rich Pepper, Djordje Petkoski, Scott Sher, and Christopher Wilson, Paul, Weiss, October 20, 2025,https://www.paulweiss.com/insights/client-memos/algorithmic-pricing-and-antitrust-risk.

15. Ibid.

16. Yaholnyk and Zeleniuk, “Antitrust Implications.”

17. Ibid.

18. Ibid.

19. Ibid.

20. Ibid.

21. Ibid.

22. Ibid. 

23.  “The Antitrust Laws,” Federal Trade Commission, June 11, 2013,https://www.ftc.gov/advice-guidance/competition-guidance/guide-antitrust-laws/antitrust-laws.

24. Ibid.

25. “Antitrust 101: Tacit Collusion,” Jeffery Amato and Tom Neuner, Winston & Strawn, December 5, 2025,https://www.winston.com/en/blogs-and-podcasts/competition-corner/antitrust-101-tacit-collusion.

26. Ibid. 

27. Ibid.

28. Maurice E. Stucke and Ariel Ezrachi, “The Role of Secondary Algorithmic Tacit Collusion in Achieving Market Alignment,” Scholarly Works, University of Tennessee College of Law, 2023,https://ir.law.utk.edu/utklaw_facpubs/979.

29. MacKay and Weinstein, “Dynamic Pricing Algorithms,” 61.

30. Ibid, 115. 

31. Ibid,111.

32. Ibid, 159.

33. Ibid, 160.

34. Ibid, 164.

35. Ibid, 165.

36. Ibid. 

37. Ibid. 

38.  Assembly Bill 325, chapter 338, 2025 California State (2025).

39. Ibid. 

40. Ibid.

41. Ibid.

42. Ibid.

43. Ibid.

44.  Senate Bill 763, chapter 426, 2025 California State (2025).

45. Ibid. 

46. Mach v. Yardi Systems, Inc., No. 24CV063117 (California Superior Court Alameda County Oct. 6, 2025).