Ten Design Guidelines To Mitigate The Risk Of AI Pricing Tool Noncompliance With The Federal Trade Commission Act, Sherman Act, And Colorado AI Act

In this article, the author explains that companies increasingly face the  risk of being targeted in lawsuits brought by governmental bodies and class  action plaintiffs alleging unfair methods of competition in violation of the  Federal Trade Commission (FTC) Act and agreements that restrain trade in  violation of the federal Sherman Act. The author identifies recently emerging  trends in such lawsuits, including one currently on appeal in the U.S. Court  of Appeals for the Third Circuit and three pending in district courts, draws  common threads, and discusses 10 guidelines that AI pricing tool designers  should consider to mitigate the risk of noncompliance with the FTC Act, the  Sherman Act, and Colorado AI Act.

"Express collusion violates antitrust law; tacit collusion does not. . . . [I]t is not a violation of antitrust law for a firm to raise its price, counting on its competitors to do likewise (but without any communication with them on the subject) and fearing the consequences if they do not."

—In re Text Messaging Antitrust Litig., 782 F.3d 867, 872, 876 (7th Cir. 2015) (Posner, J.) (case involving the Sherman Act)

While artificial intelligence (AI) pricing tools can improve revenues for retailers, suppliers, hotel operators, landlords, ridehailing platforms, airlines, ticket distributors, and more, designers and deployers of such tools increasingly face the risk of being targeted in lawsuits brought by governmental bodies and class action plaintiffs alleging unfair methods of competition in violation of the Federal Trade Commission (FTC) Act and agreements that restrain trade in violation of the federal Sherman Act.

This article identifies recently emerging trends in such lawsuits, including one currently on appeal in the U.S. Court of Appeals for the Third Circuit and three pending in district courts, draws common threads, and discusses nine guidelines that AI pricing tool designers should consider to mitigate the risk of noncompliance with the FTC Act and Sherman Act:

1. Stay tuned on FTC v. Amazon if considering allowing the algorithm to engage in tacit collusion;
2. Do not allow the algorithm to use shared nonpublic data to make individual price recommendations;
3. Do not allow the algorithm to publish customers' nonpublic information to other customers unless sufficiently nonsensitive, aggregated, and anonymized;
4. Stay tuned on the Third Circuit if considering allowing the algorithm to train with the benefit of information provided by each customer;
5. Maintain a procompetitive message to the market versus inviting a conspiracy;
6. Design and encourage pricing decision methods alternative to accepting the algorithm's recommended prices;
7. Train the algorithm with compliant pricing data;
8. Prevent algorithmic conspiracy; and
9. Audit use of the algorithm for noncompliance. As a final guideline to mitigate the risk of noncompliance with the Colorado AI Act, this article recommends:
10. Add a human between the algorithm and consumers

1. Stay Tuned on FTC v. Amazon if Considering Allowing the Algorithm to Engage in Tacit Collusion

In FTC v. Amazon.com, Inc.,¹ the FTC brought suit against a large online retailer alleging that its AI algorithm made unilateral price raises that it predicted other retailers would follow. According to the FTC, this activity, known as "tacit collusion," constituted "unfair methods of competition" in violation of the FTC Act although perfectly legal under the Sherman Act, as per this article's epigraph. The retailer moved to dismiss, arguing that unfair competition under the FTC Act requires an agreement, just like the Sherman Act requires an agreement. The FTC responded that the scope of the FTC Act reaches more broadly than the Sherman Act to cover tacit collusion by AI, thus pursuing its recently more aggressive stance regarding "unfair methods of competition" in the age of AI.²

The district court agreed with the FTC, finding that allegations of tacit collusion coupled with allegations of "anticompetitive intent or purpose" sufficed to state a claim for unfair competition under the FTC Act for purposes of Rule 12(b)(6).³

Fact discovery is proceeding, a bench trial is set for October 13, 2026, and appeals may follow. A final judgment may determine for the first time whether there exists any scope of prohibited unfair competition under the FTC Act beyond the prohibitions established by the Sherman Act. According to the retailer, the FTC cited "no case in which any district court has ever held a defendant liable on such a 'standalone' unfair method of competition claim."⁴

At stake in this heavyweight battle is the legality of a primary design choice for any AI pricing tool designer before starting to implement most or all of the design—whether to allow tacit collusion.

The remainder of this article provides guidelines to mitigate the risk of violating the Sherman Act, as well as the FTC Act, but only to the extent the FTC Act overlaps with the Sherman Act (or to the extent the AI pricing tool relies on Amazon's pricing data, see guideline  7). To the extent the FTC Act reaches beyond the Sherman Act regarding AI pricing tools, that is a new issue raised by FTC v. Amazon, so keep your eyes on it and factor it into your organization's risk management program if considering allowing the algorithm to engage in tacit collusion.

2. Do Not Allow the Algorithm to Use Shared Nonpublic Data to Make Individual Price Recommendations

Courts presiding over two AI pricing tool cases pending in district courts are drawing a bright line by prohibiting AI tools' use of competitors' nonpublic information, as illustrated by the following two cases involving AI pricing tools where the AI pricing tool vendors' motions to dismiss were denied primarily on this basis. A key point for AI algorithm designers to note here is that although shared data cannot be used for any individual price recommendations, as illustrated in this section, that prohibition does not mean that the AI pricing tool cannot train with the benefit of information provided by each customer (a separate issue discussed in guideline 4).

In In re RealPage, Inc., Rental Software Antitrust Litigation,⁵ tenants brought class actions against an AI tool vendor and its landlord customers, alleging that the vendor facilitated a pricefixing agreement by providing its customers price recommendations based on the customers' collective nonpublic "pricing and supply data," in violation of the Sherman Act. The vendor moved to dismiss, arguing that any competitor data a customer had access to was aggregated and anonymized. The district court ruled for the tenants, finding sufficient allegations nonetheless to allow the case to proceed to discovery due to the algorithm's use of "shared" nonpublic information in making its price recommendations. As the district court explained, the "most compelling evidence of horizontal agreement are allegations that [the landlord customers of the vendor] submitted real-time pricing and supply data to be compiled into a common algorithm, which was sent to all [other customers] as 'forward-looking, unit-specific pricing and supply recommendations based on their shared data' to achieve higher prices."

In Duffy v. Yardi Sys., Inc.,⁶ tenants brought class actions against an AI tool vendor and its landlord customers, alleging that the vendor facilitated a price-fixing agreement by providing its customers price recommendations based on the customers' collective nonpublic "pricing, inventory, and market data," in violation of the Sherman Act—as in In re RealPage, but against a different AI tool vendor. Also as in In re RealPage, the district court held that the tenants plausibly alleged a conspiracy in violation of the Sherman Act and found the algorithm's use of nonpublic information compelling in this regard, stating: "Defendants would have the Court assume that the lessor defendants, having turned over their commercially-sensitive data and paid for the services [the AI tool vendor] offered, did not intend to use the information generated as a result. . . . The Court finds that plaintiffs have plausibly alleged a conspiracy in violation of § 1 of the Sherman Act."

In sum, the Sherman Act prohibits AI pricing tools from using shared nonpublic data to make individual price recommendations.

3. Do Not Allow the Algorithm to Publish Customers' Nonpublic Information to Other Customers Unless Sufficiently Nonsensitive, Aggregated, and Anonymized

The Sherman Act prohibits AI pricing tools from publishing sensitive nonpublic data among its customers. Courts have upheld claims under the Sherman Act when competitors' nonpublic data were published to other competitors, unless the information was sufficiently nonsensitive, anonymized, and aggregated. For example, one court dismissed a Sherman Act claim involving exchange of anonymized and aggregated sales, production, and inventory data "(but never price data)."⁷ By contrast, other courts have allowed Sherman Act claims to go forward where statistical reports provided "access to otherwise private information on the production and prices of other Defendants" and the ability to "reverse engineer the reports to identify which Defendant provided a given data set,"⁸ or where competitors were permitted to exchange nonpublic employee compensation and budget data⁹ or the "most recent price charged or quoted."¹⁰

4. Stay Tuned on the Third Circuit if Considering Allowing the Algorithm to Train with the Benefit of Information Provided by Each Customer

A case pending in the Third Circuit involving AI pricing tools dismissed claims involving not any exchange of nonpublic information or data pooling among customers but provision by each customer of "its current, non-public . . . pricing and [inventory] data to the [AI pricing tool] platform  . . . the same third-party algorithm platform to which their co-defendants were submitting their own respective real-time and non-public pricing and [inventory] data."¹¹

Originally published by Fastcase.

To view the full article click here.

Disclaimer: This Alert has been prepared and published for informational purposes only and is not offered, nor should be construed, as legal advice. For more information, please see the firm's full disclaimer.