Introduction and Summary

On January 27, 2026, the European Commission opened a specification proceeding in connection with its designation of Google as a gatekeeper and its search platform as a core platform service into how the company should comply with Article 6(11) of the Digital Markets Act (DMA), which requires Google to share ranking, query, click and view data with rival search engines on fair, reasonable and non-discriminatory (FRAND) terms.[1] Earlier this month, the Commission issued proposed measures that Google must undertake to be in compliance with Article 6(11) that encompass eligibility, data scope, anonymization, pricing terms, and the process for data acquisition and pre-acquisition data testing.[2]

The Information Technology and Innovation Foundation (ITIF), an independent nonprofit, nonpartisan research and educational institute focusing on the intersection of technological innovation and public policy, welcomes the opportunity to respond to the Commission’s consultation on these proposed measures, where it seeks views on the proposed measures’ “effectiveness, completeness, and implementation timelines.”[3] In several respects, ITIF submits that the Commission’s proposed measures go well beyond what should be necessary for Google to comply with the DMA and will harm consumers and chill innovation in search.

ITIF’s comments mirror the structure of the proposed measures: First, ITIF explains why the eligibility criteria should not require Google to share data with artificial intelligence (AI) chatbots; second, ITIF highlights how the scope and conditions of the Commission’s proposed data requirements problematically go much further than those contemplated in the United States; third, ITIF argues that the measures necessary to protect consumer privacy would put a prohibitive compliance burden on Google; fourth, ITIF notes that the Commission’s model for FRAND pricing insufficiently protects Google’s incentives to continue to invest and innovate, and is even discriminatory in nature. Recommendations and a brief conclusion follow.  

1. Eligibility

The Commission’s proposed measures explain that, for purposes of defining the other online search engines (OSEs) with whom Google must share its data, Google “shall not exclude third-party undertakings, such as undertakings providing AI chatbots with OSE functionalities, to the extent that these undertakings provide an OSE meeting the definition set out in Article 2(5) of Regulation (EU) 2019/1150, even if the OSE is provided as part of a broader service.”[4] In so doing, the proposed measures are similar to those remedies recently imposed by U.S. courts in the United States v. Google search case, which, in fashioning their own data sharing relief, made clear that AI chatbots qualify as competitors for purposes of getting access to Google’s search data.[5] Like the intended relief in the United States, however, the Commission’s proposed data sharing measures also contain serious flaws that counsel against their implementation.[6]

To be sure, AI chatbots are pro-competitively disrupting the search market in a way that is consistent with the Schumpeterian “leapfrog” competition that has defined so many digital markets. This was recognized by none other than the very court that imposed data sharing remedies on Google in the United States, which stated in no uncertain terms that “the emergence of GenAI changed the course of this case” and was “front and center as a nascent competitive threat.”[7] In fact, OpenAI’s chief financial officer recently stated that AI is “blowing open the search markets,” noting that the company doubled its search share from roughly 6 percent to 12 percent in just six months.[8] Indeed, the numbers speak for themselves: In 2024, Google’s own U.S. market share fell consistently below 90 percent for the first time since 2015, with the traditional search market predicted to shrink by 25 percent by 2026.[9]

As such, AI chatbots do not represent competition in the market in a way that would fall under the search services that have been designated as a core platform service under the DMA, but instead reflect competition for and outside of the search market and thus beyond the scope of the DMA’s focus. Antitrust relief may sometimes restrict conduct in ancillary markets, but the DMA is a regulatory regime, not standard competition enforcement. By treating AI chatbots as being part of the same market as search, the Commission adopts an inconsistent position: If AI chatbots compete closely with search, it’s unclear where the market failure needed to justify extensive DMA-style regulation of search. Indeed, that the AI chatbot market is clearly flourishing confirms that extending the DMA into this space is unnecessary to ensure that this disruptive leapfrog competition from AI on the search space continues.

2. data scope

Article 6(11) mandates that Google must share with competitors “ranking, query, click and view data in relation to free and paid search generated by end users on its online search engines.”[10] The proposed measures explain that this data sharing shall adhere to a “parity principle,” which requires that Google “must share all query, view click and ranking data which it collects for the purposes of optimizing its search services.”[11] Specifically, Google must “make available an API or equally effective tool to share the data with third parties, at a frequency on par with Alphabet’s own frequency of access to the same data, and for a duration that can reasonably enable third parties to optimise their online search services, which is for at least five years.”[12]

To be sure, the categories of data stipulated in the DMA that Google must share overlap in part with the data that Google was ordered to share in connection with the ongoing search litigation in the United States, which included, for example, query, ranking, search engine results page (SERP), and query interpretation data.[13] However, in several respects, the proposed measures go troublingly beyond what was required in the United States v. Google search case. In particular, U.S. courts set a cap on the number of user data disclosures that, however imperfectly, helps to mitigate concerns about search competitors’ “free riding on Google’s data.”[14] And the reason is simple: “[c]ompelling such firms to share the source of their advantage…may lessen the incentive for the monopolist, the rival, or both to invest in those economically beneficial facilities.”[15]

By contrast, the Commission’s proposed measures’ “parity principle” does little to protect against free riding and promote Google’s continued investment in the search space, which is critical amid the continuing AI disruption discussed supra. Specifically, the Commission wishes search rivals to have access to Google’s data in a way that is “on par with Alphabet’s own frequency of access to the same data”—creating a huge free riding problem that is both wholly unnecessary to achieve the competitive outcomes the Commission seeks to achieve through the DMA and which will chill continued innovation in search.[16] In so doing, not only does the Commission open the door to a huge administrative burden on Google, but it troublingly doubles down on its public-utility model for implementing the DMA in complex specification proceedings. Indeed, as ITIF has previously explained, the Commission should not interpret data sharing and interoperability measures in a way that creates, rather than protects, as-efficient competitors—the latter of which reflects the broader EU competition policy with which the DMA should cohere.[17] And yet, by requiring Google to put rivals on par with itself through the sharing of user data, the latter, unreasonable goal is what the proposed measures effectively seek to accomplish.

3. Anonymisation

In addition to promoting free riding and chilling innovation, mandating that Google share sensitive user data with rivals creates significant privacy and security risks in a way that endangers the very integrity of the search ecosystem that users have long trusted—a trust that has been essential to the massive growth of search over the decades. To address this issue, the proposed measures put forward a series of technical and contractual measures to respectively “alter the search data to reduce the likelihood of re-identification of end users to a residual level without unnecessarily degrading the quality or usefulness of the search data” and provide a “set of administrative, contractual and organisational measures that complement the technical measures to further reduce the likelihood of reidentification of end users to an insignificant level.”[18]

As U.S. courts have highlighted, safeguarding user privacy in these circumstances is much easier said than done: “Think of a search query from a user in a small town regarding a rare health condition. Even if the user's name is not included in the data, context could reveal their identity.”[19] One problem is that data sharing obligations “will be subject to privacy-enhancing techniques that will diminish their full utility”—undercutting their purported competitive benefits.[20] Indeed, U.S. courts have noted both the “loss of records when using a privacy-enhancing technique known as k-anonymity” as well as how Google’s privacy filters in response to the DMA have “resulted in the exclusion of 99% of queries.”[21]

What’s more, even if privacy protections can be crafted in a way that to some sufficient extent preserves the alleged competitive benefits that the Commission seeks, measures must still be taken to ensure that they do not impose prohibitive compliance burdens on firms. For example, as the court found in United States v. Google, a cap on disclosures, among other things, “lessens the burdens associated with implementing privacy measures that will have to be applied before disclosure occurs,” and “[c]onsultation with the Technical Committee before setting a cap is critical, however, as that determination will be informed in part by the utility of the datasets disclosed after appropriate privacy-enhancing techniques are applied.”[22]

Unfortunately, the Commission’s proposed measures contain no such safeguard regarding the burden that is placed on Google to ensure user privacy. Specifically, rather than sharing data in a handful of limited, specific instances, the proposed measures contemplate the daily sharing of hundreds of millions of Europeans sensitive data with third parties—an overwhelming amount of data to ensure has adequate privacy protections. What’s more, this burden is exacerbated by the requirements set forth by Europe’s General Data Protection Regulation (GDPR), which are much more onerous than those which are likely to obtain in the context of the Google search litigation. And, further still, the proposed measures also would require Google to ensure compliance with the various contractual commitments it will be forced to undertake, including ensuring sufficient security—here again in accordance with the GDPR—which will further add to the immense burden on the company.

It is likely to prove a burden that no company—not even one as sophisticated as Google—will be able to do perfectly at scale. Simply put, imposing privacy protections for this amount of data will inherently involve automation, which is never going to be perfect—some personally identifiable information will inevitably be shared. Of course, that could open Google up to both significant potential GDPR liability, as well tremendous reputational damage to the company. Specifically, Google has not only made monumental efforts to comply with the GDPR, but has prided itself on keeping users personal data safe and private, which is a key reason why its platform has been so successful. As such, the Commission’s proposed measures appear to create a lose-lose situation for Google: Comply with the DMA by sharing huge volumes of data that create GDPR risk, or in the alternative ensure full compliance with the GDPR but then run afoul of the DMA’s massive data sharing obligations. That is not a situation that the Commission should seek to impose on any company—again, not even Google.

4. FRAND PRICING

The proposed measures specify that “Alphabet shall provide the Search Data to eligible beneficiaries against a fair, reasonable, and non-discriminatory compensation that reflects the incremental costs incurred by Alphabet for the purpose of making such Search Data available plus a reasonable return.”[23] Only “[e]xceptional circumstances may justify an additional margin.”[24] In particular, while Google “shall not be constrained to a compensation reflecting only incremental cost and a reasonable return” in cases where the company “would not be able to cover the costs incurred for collecting the relevant data” or “when an eligible beneficiary operates at a very large scale,” or alternatively when it shares data with other gatekeepers—who are, of course, disproportionately American—this does not apply to micro, small and medium-sized entities (SMEs), who “must be exempted from paying more than the incremental costs incurred by Alphabet for the purpose of making Search Data available.”[25]

The proposed measures depart substantially from the standard approach to FRAND licensing in high-technology markets. As was explained in the important Unwired Planet v. Huawei case, FRAND should reflect a “fair market price for any market participant, to reflect the true value of the SEPs to which the licence relates and without adjustment depending on the individual characteristics of a particular market participant.”[26] This is precisely what the proposed measures’ FRAND scheme does not do. First, with respect to the fair and reasonable criteria, an incremental cost measure fails to account for the full value of Google’s data, particularly the value associated with recouping the enormous fixed-cost investments that it makes into its ecosystem. Second, with respect to non-discrimination, as noted above, the proposal expressly ensures that higher rates will be paid by disproportionately American gatekeepers—for whom Google may freely abandon the incremental cost-based standard and charge higher rates—while SMEs—for whom Google must always adhere to an incremental cost-based standard—will invariably pay a much lower rate. This sort of variance among counterparties is, of course, exactly the opposite of how the non-discrimination prong of FRAND is supposed to work.

To be sure, the proposed measures’ approach is similar to the “marginal cost” standard stipulated by the court in the United States v. Google search case, not just for the sharing of user-side data, but also for the search index and ads data that Google was required to provide to rivals. As the court argued, one of the benefits of a marginal cost measure is that it’s hard to base rates on comparable arrangements given the uniqueness of the circumstances data sharing at issue such that “a court cannot look to ordinary commercial terms to set a price.”[27] Of course, not only does that rationale not fully apply here—the DMA is regulation, not court-driven law enforcement—but the court also made clear that a marginal cost measure was appropriate given that recipients were “entitled only to a one-time disclosure of a finite, closed dataset,” greatly mitigating concerns about dampening innovation.[28] However, as discussed previously, the data sharing contemplated by the proposed measures is comprehensive and ongoing, and as such is much more akin to the syndication remedy put forward in the U.S. case, where the court expressly rejected a marginal cost standard in favor of one based on market value.[29]

Recommendations

For these reasons, ITIF has substantial concerns with the proposed measures put forward to specify Google’s data sharing requirements under Article 6(11) of the DMA and offers the following recommendations for the Commission’s strong and urgent consideration:

▪AI chatbots should be ineligible for data sharing: The AI space is not only flourishing, especially at the applications level, but is also providing significant competitive constraints on Google’s market position in search. There is no need for the DMA to encompass this dynamic and still nascent space.

▪ Avoid public-utility-style obligations: The proposed measures’ “parity principle” effectively turns Google’s data sharing requirements under the DMA into a means of trying to create equally efficient competitors, rather than protecting them. Whereas protecting as-efficient competitors is central to EU competition policy, attempting to create them is a recipe for turning the DMA into heavy-handed public-utility regulation rather than a light-touch and conduct-focused model that is consistent with the Commission’s broader competition policy model. 

▪The anonymity requirements are prohibitively burdensome: The proposed measures’ scheme of sharing large amounts of sensitive user data on a daily and record-level basis puts an almost Herculean burden on Google to adequately protect user privacy as required. While ensuring privacy protections during data sharing may have been deemed feasible by U.S. courts in the United States v. Google litigation, the scope of the user data sharing there was capped to a handful of instances and was therefore far less extensive.  

▪The proposed measures’ “FRAND” pricing scheme is FRAND in name only: The proposed measures’ general limitation of Google to incremental cost pricing and blatantly divergent treatment of different counterparties is respectively and patently at odds with the “fair and reasonable” and “non-discrimination” prongs of FRAND.

Conclusion

At bottom, the proposed measures would put forward a truly radical program of data sharing that would turn the DMA into the worst kind of heavy-handed public-utility-style regulation. It would not just intervene in areas that need no regulation, but also create an environment ripe for free riding that risks greatly chilling innovation in the search space. What’s more, not only would the scope of the data sharing it mandates place a gargantuan burden on Google to ensure that users’ privacy is protected, but its concept of FRAND would both foreclose Google from adequately recouping the investments in its ecosystem and effectively help pick winners and losers among Google’s search rivals depending on whether they are a gatekeeper or an SME.

Endnotes