Key Takeaways

  • A proposed nationwide class action alleges McDonald’s used pooled, nonpublic franchise data to coordinate menu-price recommendations.
  • McDonald’s says franchisees independently determine prices and that its artificial intelligence technology does not set menu prices.
  • The dispute could help define when shared pricing algorithms create antitrust exposure for franchisors, vendors, and independent businesses.

A federal lawsuit filed this week places McDonald’s pricing technology at the center of a developing antitrust fight: Can one algorithm coordinate prices among independently owned businesses even if those businesses retain authority over the final decision?

The proposed nationwide class action alleges that McDonald’s used machine-learning models trained on nonpublic franchise data to generate coordinated menu-price recommendations. According to Reuters, the system reportedly analyzes millions of daily transactions and produces recommendations tailored to individual locations.

The allegations remain unproven, and the case is at an early stage. McDonald’s says its franchisees set their prices independently and that artificial intelligence does not determine menu prices.

That distinction will be central to the litigation. McDonald’s operates nearly 14,000 restaurants in the United States, and about 95% are independently owned. Those franchisees operate under the same brand, but their legal independence could make the use of shared pricing technology more consequential under Section 1 of the Sherman Act.

Section 1 generally addresses agreements that restrain trade. A traditional price-fixing case might involve competitors directly agreeing on what to charge. Algorithmic-pricing cases present a less tidy arrangement. Businesses can supply commercially sensitive information to a shared system, receive recommendations based partly on competitors’ data, and then make nominally independent choices.

Discretion does not automatically resolve the antitrust issue. Regulators have argued that a common pricing system can facilitate concerted action when competitors exchange sensitive information or routinely follow shared recommendations. The Federal Trade Commission has described the basic principle bluntly: price fixing by algorithm is still price fixing.

The plaintiffs’ theory appears to depend on showing more than the existence of sophisticated analytics. They would likely need to establish that independent franchisees participated in a common arrangement, directly or indirectly, and that the system reduced meaningful competition among them. Evidence about data inputs, recommendation adoption rates, franchise agreements, communications, overrides, and corporate pressure could therefore matter.

McDonald’s, by contrast, can point to franchisee autonomy. Recommendations are not necessarily commands. A tool that forecasts local demand, food costs, promotions, or customer behavior can support ordinary business planning without creating an unlawful agreement. The legal question is where analytics ends and coordination begins.

That is not merely a restaurant-industry issue. RealPage’s rental-pricing software and Cendyn’s hotel-pricing technology have already made algorithmic pricing a prominent concern in other markets. The common thread is pooled information. When nominal competitors contribute nonpublic data to the same engine, each participant may gain visibility into market conditions that would otherwise remain decentralized.

State law is moving too. California amended its Cartwright Act in 2025 to prohibit agreements among competitors to use a common pricing algorithm and coercion intended to secure adoption of algorithmic recommendations. Companies operating nationally could therefore encounter different compliance expectations across jurisdictions, even before federal courts settle the broader Sherman Act questions.

For technology leaders, the practical lesson is not to treat pricing AI as a routine optimization deployment. Governance reviews can examine what information enters the model, whether data is aggregated or identifiable, how recommendations are communicated, and whether users face incentives or penalties tied to adoption. Audit trails can also document when human decision-makers reject or modify suggested prices.

The NIST AI Risk Management Framework offers a useful structure for mapping systems, measuring risks, and assigning oversight, although it is not an antitrust standard. Legal analysis still needs to address market relationships, data-sharing arrangements, and evidence of agreement.

What happens if franchisees are free to reject recommendations but rarely do? That question may become one of the case’s more important tests. Regardless of the eventual outcome, the McDonald’s lawsuit signals that companies deploying shared pricing engines should evaluate not only whether the model performs well, but also how its data and recommendations shape competition among the people using it.