Bridging Internet & AI Governance: From Theory to Practice

 IGF 2025, here.

Interoperability in Digital Platforms and its Regulation: Transatlantic Dialogue alive and kicking!

Not a dead parrot.

Webinar, 1 July, 15-17 CET. Register here if you want to pose questions, otherwise live on YouTube here. Recording here.

This webinar is a transatlantic conversation bringing together researchers, regulators and civil society. The focus is on interoperability, not as a silver bullet, but as a critical lever to support more open, competitive and innovative digital environments.
It offers a timely comparison: Brazil, a jurisdiction that has demonstrated its ability to build effective digital public infrastructure (Pix), thereby getting rid of the extractive Visa-Mastercard duopoly, and the European Union, which has so far struggled to do the same. At the same time, Europe has taken the lead in legislating to curb Big Tech’s power, and other regions, including Brazil, are now watching the Commission's enforcement of this legislation closely.
All this, just as transatlantic tensions over digital regulation resurface, and as the EC DMA Team does its utmost to stay below Trump’s radar. And then there's the DMCCA (and UK politics). 

Together with Isa Stasi and Ian Brown, our task on Tuesday is to share a few lessons from the European experience. So what is it, really, that we have to offer from this side of the Atlantic?

As for my contribution, I’m still finalising the details. Not long ago, I wouldn’t have had much to say about EU interoperability, at least not anything terribly useful for promoting open, fair and competitive digital markets. But the past few months have been surprisingly lively. Four developments stand out, and I hope they can add a little spice to our conversation. I will most likely begin with the antitrust commitments by Apple concerning NFC (Apple Pay), and reflect on their aftermath. Next, I’ll briefly touch on the recent judgment of the Court of Justice of the European Union regarding interoperability of the Android Auto OS. I’ll then say a few words about the Commission’s specification decisions on Apple’s interoperability obligations under Article 6(7) of the DMA. And finally, I’ll offer some thoughts on prospects for stronger DMA enforcement, on the case for refining the regulatory framework, and even on the EuroStack (10 minutes in total :-)). 

The first reflection I would like to offer concerns access to Near-Field Communication (NFC) functionality, a technology which, until mid-2024, Apple had reserved exclusively for its own Apple Pay service within the EEA. An important point to note is that, across Europe, NFC, a technology not developed by Apple, has become the standard for mobile payment. It enables fast, contactless transactions, secured through tokenisation and encryption. Virtually all payment terminals in the EEA now support it. 

It is now almost exactly one year since the European Commission made Apple’s commitments in the Apple Pay case legally binding. These commitments are centred squarely on interoperability: Apple is required to allow third parties access to the NFC functionality for payment purposes on iOS devices. As a result, a wide range of developers can, in principle, begin to use this technology to offer alternative NFC  payment services. Even though relatively little time has passed, it is important, I believe, to ask whether anyone has actually seized this opportunity, whether any new entrants have made their way into the NFC in-store mobile wallet market on iOS. There have, in fact, been some entries, though so far limited to a few countries rather than on an EU-wide scale. As illustrated at the most recent OECD Competition Committee meeting in June, the first to enter was Vipps MobilePay, though its launch remains limited to Norway, and facing huge hindrances to becoming a pan-European interoperable wallet (Single Market, anyone?). Next came a US tech firm, hardly a small player, namely PayPal, which is currently rolling out its wallet in Germany. German cooperative banks have also signalled their intention to enter this space soon, likewise focusing on the German market. In the announcement, it is explicitly stated that Apple Pay will no longer be needed to make payments with the new service, a move framed as part of a broader effort to raise awareness of how heavily payment systems in Europe rely on US corporations such as Visa, Mastercard, PayPal, and, of course, Apple. The ongoing trade tensions with the US are cited as an additional reason for concern. This raises a broader question: can interoperability serve not only as a tool to promote competition, but also as a means of advancing digital sovereignty? The answer, perhaps, is that interoperability is certainly a first step, but a far more effective approach, had it been pursued from the outset, would have been to establish a digital public infrastructure for electronic payments, along the lines of Brazil’s Pix. Crucially, this would have required a broad adoption mandate for banks operating across the EEA. If done properly, such a system could have delivered both competition and sovereignty in a more structural and sustainable way. A related and important question is what went wrong with SEPA, the Single Euro Payments Area. Conceived as a cornerstone of European financial integration, SEPA has largely failed to deliver the kind of common digital payment infrastructure that could support genuine sovereignty and competition at scale. Even though scepticism remains high, it remains to be seen whether the Instant Payments Regulation, now in force, with serious enforcement beginning this year, will offer an effective fix to SEPA’s shortcomings. The other, and perhaps enduring, question is whether the commitments offered by Apple were ever sufficient. More fundamentally, even if the commitments had been ideal, one must ask how much they could realistically achieve in isolation. If anticompetitive conditions persist in adjacent markets, and addressing interoperability at just one layer may do little to resolve distortions that are structural and multi-layered in nature (e.g., terminals?). Moreover, to fully benefit, at the EEA-scale, from access to previously gated functionalities, new entrants would need to rely on other components of essential digital public infrastructure, most notably, the European Digital Identity, meant to be enabled by the eIDAS 2.0 framework.

Even from the few observations above, it becomes clear that getting interoperability right, as a driver of innovation, competition, and even digital sovereignty, is no small feat. It requires multiple elements to come together in a coherent and sustained manner. It is far from simply unleashing latent energies held back by a textbook refusal to provide interoperability. In the Brazilian context, the national scope of the market, combined with the groundwork already laid in this area by the regulator, may well place the competition authority in a favourable position to act effectively. That said, it is beyond doubt that the refusal to enable interoperability has often been used strategically by Big Tech players as an anticompetitive tool.  In some cases, it served to block potentially disruptive innovators at a critical moment; in others, it was used to secure and preserve market positions in areas where new business opportunities were emerging, as Apple did with NFC functionality until recently, and as Google did in relation to Android Auto, for which it was sanctioned by the Italian competition authority. 

This latter case also triggered a preliminary reference to the EU Court of Justice, which did not miss the opportunity to say once again something helpful, a development I would like to briefly address as the second point in my remarks. That the effectiveness of the branch of competition law tasked with preventing and prohibiting abuses of market power has been profoundly challenged by the rise of Big Tech is, of course, no secret. One aspect long recognised as particularly ill-suited to the digital context is the so-called Essential Facilities Doctrine, particularly in its Bronner formulation.The Court’s ruling in Android Auto provides a further fix, in the form of a "clarification" of the doctrine, perhaps a limited one, but a welcome one nonetheless. The case concerned Google’s refusal to allow Enel X’s electric vehicle charging app to interoperate with the OS Android Auto, citing security concerns and the burdens of developing a new template. The Italian competition authority (ICA) found this refusal to be in breach of Article 102 TFEU and ordered Google to enable interoperability. 

Last slide of my 2021 presentation 

In a 2021 ASCOLA Conference online presentation (Covid time, my last one) discussing the case, my first research question was whether the legal and economic reasoning proposed by the ICA would have made it more agile for competition authorities to assess interoperability cases involving digital platforms under Art. 102 TFEU. We can, at this point, confidently answer in the affirmative, and that is good news. From my perspective, what deserves particular emphasis is that the Court made a meaningful contribution on the issue of safeguarding investment and innovation incentives. While the arguments put forward certainly merit further reflection, the Court’s reasoning was sufficient to effectively dismantle the familiar defence invariably deployed by Big Tech in abuse of dominance cases. Unlike the infrastructure at issue in Bronner, Google had not developed the Android Auto OS to serve its own needs to secure a competitive advantage. On the contrary, it was explicitly designed to encourage participation by complementors, a fact that clearly undermines the argument that granting access would have reduced Google’s incentives to invest. To what extent this workaround can be relied upon beyond the specific platform ("complementors") context and across different digital services and types of infrastructure remains to be seen. 

The second question I raised during the ASCOLA presentation, bearing in mind that this was back in 2021, after the DMA had been tabled but before it became law, was how to frame the relationship between ex post and ex ante approaches to interoperability mandates going forward. The ex ante approach to interoperability introduced by the DMA will form the third point of my remarks. But before that, a few words are in order on the relationship between ex ante and ex post interventions and, more broadly, between traditional antitrust law and emerging forms of digital regulation with regard to interoperability, which can be of some interest specifically in the context of our transatlantic dialogue.  The first thing to note is that the Commission’s experience in the Apple Pay commitments proceedings, discussed earlier, appears to have fed directly into the current phase of DMA enforcement. This is evident in the confidence with which the Commission has now moved to specify Apple’s interoperability obligations under Article 6(7), a point I will return to shortly. The second aspect I wish to highlight, though much more could be said, is that courts are reading the DMA and drawing inspiration from it, even when interpreting traditional antitrust law. This is clearly visible in the Android Auto case, where a regulatory approach inspired by the DMA can be seen in how the Court assessed what may constitute an objective justification for refusing to grant interoperability. In my view, all things considered, this too is a positive development.

The prince app developer (see blog post)

With the third point, we turn to the DMA. Ensuring certain levels of interoperability is clearly a priority for the EU legislator, as a means to promote digital markets that are fair and contestable. This is evident from a range of obligations whose proper compliance by gatekeepers presupposes different degrees of interoperability between their services and third party services. In this context, I would like to say a few words about Article 6(7), which has recently been the focus of significant enforcement activity by the Commission. As I write these lines, the 2025 Apple Enforcement Workshop is taking place, with an entire session dedicated to Apple’s compliance journey with this very provision. Apple was required to comply with Article 6(7), and the other DMA obligations, by 7 March 2024. This has evidently not occurred, and the Commission has taken notice. To address such situations, the DMA introduces a new tool: the specification proceeding. Under Article 8, the Commission can unilaterally determine the measures needed for compliance, unlike commitment proceedings, which rely on the company’s own proposals to avoid a formal finding of antitrust infringement. The Commission issued two specification decisions in relation to Apple last March: one concerning the process for handling interoperability requests, and the other relating to interoperability with connected devices. With the latter, the Commission specifies, among other things, how Apple should provide effective interoperability with the NFC controller in Reader/Writer mode by the end of 2025 at the latest. The scope of the NFC part of the decision is therefore significantly broader than last year’s antitrust commitments. It is also clear that interoperability with this feature is in high demand and, in principle, should foster innovation. This is reinforced by the fact that Article 6(7) does not allow Apple to dictate which services or products may make use of the interoperability obligation. What stands out in this respect is also the in-house technological expertise the Commission has already developed regarding the services offered by the gatekeepers it regulates. This expertise enables it, as in this case, to engage eye to eye with a giant like Apple.

The fourth and final aspect I would like to briefly touch on relates to a pronounced sense of urgency. For instance, I find it worrying that it is considered normal for a study on the impact of emerging technologies on the DMA to take up to a year to produce. In the same vein, if we come to realise that the DMA is insufficient, whether in terms of its scope of core platform services or the need for additional obligations, then the time to act is now, and this relates also to the possible need for additional interoperability obligations. We cannot afford to lose time. However, even for the most determined regulator (and well equipped, also financially) in the world, it is very difficult, though not impossible, as we have just seen, to impose interoperability obligations on powerful and recalcitrant gatekeepers. But that is precisely what lies ahead if we allow them to build the digital infrastructure on which we all then come to depend. Equally difficult, and arguably wishful thinking, is relying on the industry to deliver effective interoperability solutions. This has been attempted in the EU for years, with very limited results. So how do we move forward? Well, perhaps by starting with a good look back. We need retrospective analyses to understand how we ended up letting a handful of players control the digital infrastructures we all rely on, not to point fingers at antitrust enforcement alone, but to get a clearer picture of the broader dynamics, which likely vary across different digital services. On this, as on many other things, I fully agree with Robin Berjon, who notes that everything in digital happened so quickly we didn’t even realise we were dealing with infrastructures, let alone how to properly regulate them. Secondly, the kind of public effort required, in terms of both capacity-building for smart regulation and "project" funding, must be laser-focused on identifying the type of infrastructures we need and how to get there. This type of mobilisation cannot be decoupled from public choices. This is particularly clear in the case of Pix, where inclusiveness was a key consideration (but almost accidentally ended up by solving an extractive duopoly problem). Moreover, infrastructural choices inevitably shape the kinds of technological innovation they enable. These are societal decisions, not technical footnotes (and, of course, we cannot afford for them to be significantly steered by even well-meaning think tanks nor other - European! - industry giants, from telecoms to aerospace). The upcoming plenary debate in the European Parliament on the Report on European Technological Sovereignty and Digital Infrastructure, produced by the Committee on Industry, Research and Energy (ITRE), offers an excellent starting point for these much-needed democratic conversations.

To be discussed!

 

Workshops Reveal the DMA’s Broken Promises

 Chamber of Progress, here.

Facebook is starting to feed its AI with private, unpublished photos

The Verge, here.

After Notice and Choice: Reinvigorating “Unfairness” to Rein In Data Abuses

 L. Khan, S. Levine, S. Nguyen, here.

First-sale doctrine in the AI age: incentivising book bonfires! No externalities there?

 ArsTechnica, here. 

Picture chosen by ArsTechnica for the article.

This picture suddenly recalled a Ted Talk I gave 12 years ago touching also on the first sale doctrine.

Unbelievable how much went wrong since.