Contents

1. Introduction and Summary 1

2. Additional Guidance Topics 2

3. New Technologies 4

4. Significant Developments 5

Recommendations 6

Conclusion. 7

Endnotes 7

1. Introduction and Summary

On February 23, 2026, the Department of Justice (DOJ) and the Federal Trade Commission (FTC) issued an RFI Notice regarding potential additional guidance on collaborations among competitors (Notice).[1] The Notice follows the withdrawal of the previously longstanding 2000 Antitrust Guidelines for Collaborations Among Competitors (2000 Guidelines).[2] The Biden administration withdrew the 2000 Guidelines at the very end of its term, claiming at the time that they “no longer provide reliable guidance to the public about how enforcers assess the legality of collaborations involving competitors.”[3]

The Biden administration offered four reasons for withdrawing the 2000 Guidelines: First, they did not adequately reflect decades of subsequent case law; second, they relied “in part on outdated and withdrawn policy statements and risk creating safe harbors that have no basis in federal antitrust statutes”; third, they “rely on outdated analytical methods that fail to capture advances in computer science, business strategy, and economic disciplines”; and fourth, that they “fail to address the competitive implications of modern business combinations and rapidly changing technologies such as artificial intelligence, algorithmic pricing models, vertical integration, and roll ups.”[4]

The Information Technology and Innovation Foundation’s (ITIF) Schumpeter Project on Competition Policy greatly appreciates the opportunity to comment on the Notice and agrees with the DOJ and FTC on the need for guidance in this area to “provide business with the predictability and confidence they need to collaborate and grow.”[5] It was a costly error for the Biden administration to withdraw the 2000 Guidelines without replacing them. Issuing new and revised guidelines that reflect sound legal and economic principles—particularly for assessing the competitive merits of horizontal agreements between competitors—is therefore timely and necessary.

This comment addresses the three questions posed by the Notice. First, ITIF presents a number of topics that might benefit from additional guidance, especially the need for safe harbors to provide certainty to the business community; second, ITIF discusses new technologies and business models that would benefit from additional guidance, in particular, artificial intelligence (AI) and related issues of so-called “algorithmic collusion”; third, ITIF highlights several developments that “should be considered in any revisions to the prior competitor collaboration guidelines,” including significant legal, economic, or technological developments, as well as specific key Supreme Court cases that should inform any future guidance.[6] Recommendations and a brief conclusion follow.

2. Additional Guidance Topics

Safe harbors are an excellent way to achieve administrability benefits and reduce compliance costs for common and procompetitive business conduct. Unfortunately, by withdrawing the 2000 Guidelines, the Biden administration eliminated important safe harbors or “safety zones” that existed both for competitor collaborations in general as well as research and development (R&D) joint ventures in particular.[7] In addition, the Biden administration withdrew several longstanding policy statements in the healthcare space that included safe harbors in the area of information sharing.[8] That withdrawal broadly affected many stakeholders, which had treated the information-sharing safety zone framework as generally applicable.[9]

2.1 General Safe Harbor

Even more so than mergers, competitor collaborations are ubiquitous across the economy and only in very rare circumstances create competitive concerns. For this reason, the 2000 Guidelines rightly included a structural presumption of legality for competitor collaborations evaluated under the rule of reason “when the market shares of the collaboration and its participants collectively account for no more than twenty percent of each relevant market in which competition may be affected.”[10] This presumption should be restored in future guidelines: Agreements analyzed under the rule of reason are broadly incapable of harming competition absent market power, which is highly unlikely when the parties’ combined market shares fall below 20 percent. 

2.2 R&D Joint Ventures

The risk of false positives is particularly acute in assessing competitor collaborations that facilitate innovation, a fundamental dimensionality of competition and the greatest long-run driver of economic growth. Accordingly, the 2000 Guidelines made clear that R&D joint ventures would be presumptively lawful “where three or more independently controlled research efforts in addition to those of the collaboration possess the specialized assets or characteristics and the incentive to engage in R&D that is a close substitute for the R&D activity of the collaboration.”[11] While the general causal relationship between market structure and innovation remains a matter of debate, consistent with the inverted-U literature that has blossomed since the 2000 Guidelines, new guidelines should consider reiterating a structural presumption of legality for R&D joint ventures which fall short of monopoly or duopoly, and thus any meaningful possibility of harmful unilateral or coordinated anticompetitive effects in an innovation market.[12]

2.3 Information Sharing

Information sharing between competitors is another specific practice that is widespread and overwhelmingly procompetitive in the modern economy. Prior to the Biden administration, the DOJ and FTC had long adhered to a sensible safe harbor for exchanges that did not pose a material risk of harming competition: First, “the collection is managed by a third party”; second, “the information provided by participants is more than three months old”; and third, “at least five participants provide the data underlying each statistic shared, no single provider’s data contributes more than 25% of the ‘weight’ of any statistic shared, and the shared statistics are sufficiently aggregated that no participant can discern the data of any other participant.”[13]

There were no legal or economic developments that justified the Biden administration’s decision to terminate this safe harbor. In general, information exchanges are only likely to result in harm to competition when they involve the exchange of competitively sensitive information, which is primarily determined by three factors: First, the extent to which the information is strategic or instead routine; second, the extent to which the information is historical or, by contrast, timely; and third, the extent to which the information is aggregated and anonymized, as opposed to company-specific. The exchange of information that is not strategic, sufficiently historical, and appropriately aggregated should be presumptively lawful.

The Agencies should issue updated guidelines establishing a clear safe harbor for information sharing, which would provide significant administrability benefits that far outweigh any harm from false negatives. As a first step, the Agencies should restore the previous safety zone from the 1996 Statements of Antitrust Enforcement Policy in Healthcare as a baseline safe harbor for information sharing agreements.[14] That baseline, however, should not be the endpoint. The Agencies should build on it by adopting a tiered framework that reflects the realities of the modern data-driven economy and encourages greater use of non-competitively sensitive data. Properly designed, this approach would expand safe harbors for clearly benign and procompetitive forms of data sharing—such as aggregated, anonymized, or sufficiently lagged datasets—while maintaining appropriate scrutiny for exchanges that carry higher coordination risks. In this way, the Agencies can move beyond a static, restrictive model toward a more dynamic regime that both preserves competition safeguards and facilitates the efficiency gains associated with responsible data sharing.

3. New Technologies

Just as the 2000 Guidelines were issued at the beginning of the Internet revolution, so too are new guidelines likely to coincide with the continued boom in AI—the next great technological wave. Indeed, while some have argued that AI algorithms may make it easier for firms to form and maintain cartel agreements, it is far too early to suggest that the antitrust laws are not up to the task.[15] However, the Agencies can use the opportunity presented by potential new guidelines to provide clarity on how they will view competitor collaborations involving AI algorithms, and in particular, given the Trump administration’s continued interest in the “proper legal framework for analyzing claims involving the joint use of algorithms.”[16]

3.1 Algorithmic Collusion

Across several statements of interest and briefs involving algorithmic collusion cases, the Biden administration took the view that there is a per se violation when “competitors knowingly combine their sensitive, nonpublic pricing and supply information in an algorithm that they rely upon in making pricing decisions, with the knowledge and expectation that other competitors will do the same.”[17] The Biden DOJ argued that, to circumstantially prove a per se illegal agreement based on the use of common algorithms, there was no need to demonstrate some form of conscious parallelism, claiming that it is wrong “to suggest that a complaint must allege a binding enforcement mechanism to state a valid claim for per se price fixing,” for example.[18]

Courts have rejected the DOJ’s view. In the RealPage case, the court declined to infer the existence of a per se agreement, finding that the pricing recommendations were not followed as much as 10 to 20 percent of the time.[19] And, as the Ninth Circuit made clear in Gibson, even assuming conscious parallelism—e.g., that the algorithm was followed over 90 percent of the time—the common use of an algorithm is not enough to infer a per se illegal agreement, as each competitor independently using the same algorithm is not the same as an agreement between them to use the algorithm or use its prices as starting points.[20]

To be sure, binary “acceptance rates are not necessarily the best measure of its influence,” whereby “widening the definition of acceptance even slightly to account for partial acceptances illustrates the influence of recommendations.”[21] Accordingly, rather than relying on rigid quantitative thresholds, the Agencies should clarify that high levels of alignment with algorithmic recommendations, standing alone, are insufficient to support a per se inference of agreement. Any safe harbor in this area should instead focus on indicia of independent decision-making—such as the absence of competitively sensitive data sharing, joint optimization, or mechanisms that constrain firms’ ability to deviate—thereby avoiding distortions to the design and use of algorithmic pricing tools.

3.2 Section 5 of the FTC Act

Some have raised concerns that the independent use of algorithms could result in tacit collusion that harms consumers without being actionable under the Sherman Act.[22] But rather than attempt to distort the requirements of Section 1’s agreement requirement, the Agencies’ focus should be on utilizing Section 5 of the FTC Act to police anticompetitive invitations to collude and unilateral facilitating practices achieved through the use of algorithms. As courts have held, unilateral facilitating practices “may be labelled ‘unfair’ within the meaning of § 5 a minimum standard demands that, absent a tacit agreement, at least some indicia of oppressiveness must exist such as (1) evidence of anticompetitive intent or purpose on the part of the producer charged, or (2) the absence of an independent legitimate business reason for its conduct.”[23]

4. Significant Developments

ITIF broadly agrees with the general principles and analytical framework for assessing competitor collaborations set forth in the 2000 Guidelines. Collaborations should be assessed on whether they “harm competition and consumers,” where the specific rule applied will vary with the nature of the agreement.[24] The Agencies must also assess “the overall collaboration,” which thus accounts for the bedrock ancillary restraints doctrine.[25] However, several subsequent Supreme Court decisions present bases for modifications in any future guidelines, and in particular, the landmark decisions issued in Ohio v. Amex and NCAA v. Alston.

4.1 Platform Markets

In presenting the framework for analyzing agreements under the rule of reason, the 2000 Guidelines provided descriptions of goods markets, technology markets, and innovation markets. But any future guidelines must take note of the special considerations that apply to platform markets as described in Amex—which are generally applicable beyond the vertical restraints involved in that case. As the Supreme Court made clear, courts must consider both sides of a transaction platform when defining a relevant market in these cases, given that, as even compared to other two-sided platforms, “[t]ransaction platforms are thus better understood as ‘suppl[ying] only one product’—transactions.”[26] Doing so ensures that market definition does not result in false positives for platform markets that are often dynamic and driven by innovation competition.

4.2 Labor Markets

Since the 2000 Guidelines, many have suggested that, beyond exemptions for labor unions, antitrust should play an even greater role in protecting labor.[27] These sentiments were given life by the Supreme Court’s decision in NCAA v. Alston, where it found that the NCAA’s restraints were anticompetitive under the rule of reason.[28] That being said, there is no need for new guidelines to include any distinct section on labor market harms. This is because harm to labor, by itself, is not cognizable antitrust harm: the antitrust laws protect the welfare of consumers in a product market, not sellers in a labor market. Indeed, competitor collaborations may often involve tradeoffs whereby consumer welfare is enhanced but labor rents are reduced.

To be sure, there are two cases where conduct that harms labor is also anticompetitive due to a reduction in labor output that results in consumer harm. The first of these involves, as in Alston, “monopsony control in the relevant market…restricting the quantity of student-athlete labor,” where the relevant market is consumer-facing—foreclosing any need to define a separate labor market.[29] The second are cases involving an employer cartel. But in this instance, as with any per se violation, there is of course no need to define a relevant market of any kind, making specific labor market guidance also unnecessary in any future competitor collaboration guidelines.

4.3 Quick Look

The Alston case presents another angle for considering how to craft any future competitor collaboration guidelines. The 2000 Guidelines broadly present a binary framework between the per se rule and the rule of reason in analyzing competitor collaborations. To be sure, at about the same time as the 2000 Guidelines were released, the Supreme Court issued its opinion in Cal Dental, where it expressly embraced the “quick look” analysis it had utilized in prior cases.[30] Indeed, the 2000 Guidelines tacitly acknowledge the quick look approach when they note:

…where the likelihood of anticompetitive harm is evident from the nature of the agreement, or anticompetitive harm has resulted from an agreement already in operation, then, absent overriding benefits that could offset the anticompetitive harm, the Agencies challenge such agreements without a detailed market analysis.[31]

Future guidelines should more clearly distinguish between the per se, quick look, and rule of reason modes of analysis. Indeed, this is particularly important in light of the Supreme Court’s decision in Alston and its explicit recognition of a rule of “quick-look” approval in certain cases where the restraint does not involve any market power.[32] Specifically, just as certain collaborations may be condemned by virtue of an inspection of either their purpose or direct evidence of anticompetitive effects, so too should certain collaborations be presumed lawful due to their inherently beneficial purpose or direct evidence of net consumer benefits.

Recommendations

For these reasons, ITIF is optimistic that the DOJ and the FTC have an opportunity to produce new and improved competitor collaboration guidelines that account for the following considerations:

▪ Safe harbors are critical: Given the widespread and procompetitive nature of competitor collaborations, a general safe harbor, in addition to specific safe harbors in the areas of R&D joint ventures, which will safeguard investment and innovation, and information sharing, which ensures companies have access to useful and timely data, provides significant administrability benefits that far outweigh any possible costs from false negatives.

▪ A course correction is needed for algorithms:In view of the Agencies’ continued focus on competitors’ joint use of algorithms, future competitor collaboration guidelines provide a great avenue to make clear that a per se unlawful agreement will only be circumstantially inferred in cases where there is true parallel pricing behavior, and that companies may even be able to benefit from a safe harbor if they can show that the requisite degree of parallelism does not exist.

▪ Amex and Alston represent major legal developments: The Supreme Court’s landmark decisions in Amex and Alston made clear both that special considerations must apply when analyzing agreements that involve transaction platforms to mitigate the risk of false positives and that a “quick-look” approval rule is available to condone certain procompetitive competitor collaborations where market power has not been demonstrated.

Conclusion

While ITIF does not support the Biden administration’s decision to withdraw and not replace the 2000 Guidelines, now, a quarter century after their issuance, it is long past time for renewed competitor collaboration guidelines. Like mergers—if not far more so—competitor collaborations are not only generally procompetitive but ubiquitous throughout the economy, making the need for clear and administrable guidance paramount to provide businesses with the requisite certainty they need to compete while complying with an antitrust regime that protects competition and consumers.

Endnotes