Navigating Antitrust Scrutiny of Algorithmic Software

Federal and state antitrust enforcers are sending a clear signal to companies: An algorithm is not a shield for anticompetitive conduct. As algorithmic pricing becomes a primary focus for regulators, particularly in California, companies must prepare for aggressive enforcement.

Algorithms as the new frontier of conspiracies to restrain trade

Federal and state authorities are prioritizing enforcement actions where software platforms act as a “hub” in a hub-and-spoke conspiracy, allegedly allowing competitors to exchange competitively sensitive information and align prices without direct communication. Two recent settlements highlight the type of conduct likely to garner scrutiny from the antitrust authorities:

• United States v. RealPage: In November 2025, the Department of Justice (DOJ) reached a landmark settlement with RealPage to resolve claims that its revenue management software facilitated an algorithmic information-sharing conspiracy among competing landlords. Under the proposed settlement agreement, RealPage must stop, inter alia, utilizing rivals’ nonpublic, competitively sensitive data to generate real-time rental price recommendations and remove or modify product features, such as “auto-accept” defaults, that steer users toward aligned pricing or competitive terms.
• United States et al. v. Agri Stats, Inc.: In May 2026, the DOJ and a coalition of States reached a settlement in the Agri Stats antitrust case challenging a data analytics firm’s provision of reports containing price, output, and cost information to competing meat processors. The proposed settlement prohibits the reporting of nonpublic pricing information, granular metrics, participant identities and competitor rankings, while enforcing strict age limits on surviving historical data and requiring that remaining reports be made transparently available to all domestic purchasers on equal terms.

State enforcement: California leading the charge

California’s AB 325 – The Cartwright Act

The States continue to take aggressive action to fill a perceived gap left by the federal government when it comes to antitrust regulation and enforcement. Effective January 1, 2026, California’s AB 325 set a new national benchmark for algorithmic regulation by expressly prohibiting certain pricing algorithms. While other states, like New York, have enacted algorithmic pricing bans in certain industries (real estate) and disclosure requirements, California has gone the farthest in policing algorithmic pricing tools.

• Ban on use of algorithms to restrain trade: The law makes explicit that it is unlawful under the Cartwright Act to use or distribute a “common pricing algorithm” as part of an agreement to restrain trade or fix prices.
• Coercion focus: The law prohibits one party from “coerc[ing]” another to set a price or commercial term recommended by a common pricing algorithm. “Coercion” is not defined in the statute; at a recent conference, speakers suggested algorithms with auto-populating or auto-accepting features could be viewed as forms of coercion.
• Broad scope: A pricing algorithm is considered “common” if it has two or more users and uses competitor data to recommend or influence prices or commercial terms.

While California has yet to bring a suit under AB 325, at a recent conference, California’s Senior Assistant Attorney General for Antitrust Paula Blizzard indicated her view that:

• The focus is on the “coercion” prong of the statute.
• “Competitor data” as used in AB 325 includes any competitor data, even publicly available information.

Practical compliance checklist

To mitigate risk in this high-scrutiny environment, firms should consider the following practical compliance steps (these may differ depending on whether the firm is a developer or user of the algorithm):

• Audit algorithmic inputs: Assess the data that is input and used to train your algorithmic tools. Do the tools utilize competitor data? Is the data proprietary or publicly scraped?
• Ensure users have free choice: Avoid penalizing partners that don’t use pricing features or rewarding those that do. Consider avoiding “auto-accept” or “auto-populate” pricing features and instead ensure implementation authority requires independent human decision-making.
• Who has access? Are the algorithmic recommendations available broadly to anyone in the industry or only offered to one side of the transaction (e.g., suppliers versus customers)?
• Who else is using the algorithm? What do you know about who else is using the algorithm? For example, be careful about marketing materials that indicate the algorithmic software is used industrywide or by all competitors, or that the value of the tool can be obtained only through broad adoption.
• Discovery awareness: Enforcers are increasingly targeting AI prompts and log information. Treat all prompts and log information entered into AI agents as discoverable material, similar to executive emails. Document the “procompetitive” benefits of tools where appropriate and accurate – e.g., efficiently matching supply and demand to increase output in competition with others.