Bruce Lawson's personal site : Google’s second DMA compliance workshop: Hotseat cold feet

Google’s second DMA compliance workshop: Hotseat cold feet

Tuesday 8 July 2025

(Last Updated on 4 August 2025)

Last week, I attended the second DMA compliance workshop by Alphabet, where Google representatives explained what it’s done in the year since the DMA came into force. Last year, Google put a good face on, bringing technical staff from the Android and Chrome teams, and saying “we absolutely intend to comply with the DMA”.

This year, it was lawyers and policy people. Like Microsoft and Apple, the struggling Mountain View start-up immediately complained about the law. Claire Kelly, who leads Google’s competition team responsible for compliance, set the scene immediately:

we remain genuinely concerned about real-world consequences of the DMA, which are leading to worse online products and experiences for Europeans.

Although the DMA strives for fairness, we believe that it inadvertently curbs innovation by deterring investment in novel features and services, effectively preventing the latest technological advancements from reaching European consumers and fostering a second-class digital citizen experience compared to other regions. This happens when we stop focusing on who matters, the EU consumer, and instead listen to those with vested interests that shout loudest.

This is directly from the leaked Google lobbying strategy, one plank of which is to re-frame the political narrative around costs to the economy and consumers.

So, onto my own vested interest: smaller browsers that have to compete with the Operating System’s default.

Browser choice screen

In her introductory remarks, Claire from Google made this ominous statement.

Gatekeepers must comply with the letter of the law.

I foolishly believed that Google would aim to comply with the letter and the spirit of the DMA, given that it’s probably the most open of the Gatekeepers, both technically and culturally.

Claire described the browser choice screen:

The browser choice screen on Android is not shown if there’s a third party browser default. And likewise, the search engine choice screen in Chrome is not shown if a third party search engine is set as the default. The choice screens have forced scrolling, which requires a user to go through all options before they exercise any choice.

James Heppell from Open Web Advocacy asked why Google was not replacing the Chrome logo in the “hot seat” with the user’s default browser when they decided to change it:

We heard yesterday that Apple’s choice screen implementation now gives the chosen browser the hot seat, meaning that it’s placed prominently at the center of the dock on the home screen. replacing the Gatekeeper browser. In contrast, Google has not made these changes and continues to use a far less effective design.

We know from our conversations with browser vendors that having the hot seat leads to higher retention rates and reduces the dominance of the operating system, Gatekeeper and their browser. By not placing the user’s chosen browser in the hot seat, Google is undermining Article 6(3) and making the choice screen significantly less effective. Just setting the default browser is not enough to nullify the advantage that Google gives their own browser via their control of Android’s default setup.

Given that Google already benefits from the hot seat on iOS, and we have heard Google mention this about the hot seat, our question is, can Google commit to implementing an equivalent hot seat placement for browsers on Android, and if so, when can we expect that to roll out?

Claire responded:

Again, our approach to this is informed by what the DMA says. And Article 6.3 is clearly about defaults. A Hot Seat is not a default. So a default is a service that will trigger as a result of a generic user action, as I mentioned earlier. And so if a user goes to do something, for example, clicks on a URL, it would be the browser that is used in order to then fulfill that specific user intents.

The placement of a particular app on a device has nothing to do with the default. And so that’s our view in respect of how the hot seat question fits with what we’re talking about when we talk about compliance with the DMA.

Kush Amlani, Director of Global Competition and Regulation for Mozilla, commented

coming back to the hot seat and to your response there, I wanted to say that if you select a browser via a choice screen and it downloads to page 13 of your apps I’m not sure that necessarily complies with 6.3 and would contest your interpretation.

Recital 49 of the DMA says that gatekeepers should allow end users to easily change default settings when those default settings favour their own software applications and services. And Article 6.3 also says that the gatekeeper should allow and technically enable end users to easily change default settings, including by prompting via a choice screen.

So from our point of view, we think both of those things do require you to place the selected browser in the hot seat. And obviously it’s already been done by other gatekeepers as well.

I agree with Kush, and not only because he’s bigger than me. But, perhaps unsurprisingly, Claire doesn’t, and had not changed her mind over the course of the conversation. Google would stick to the letter of the law, and even out-stickler Apple (in this matter, at least):

we think our position on this one is is pretty clear in respect of the hot seat, I mean, again, we can understand your perspective on things, but again, it has to come back to what’s said in the DMA.

Even your reference to the recital refers to default. Default means a very, very specific thing, and it hasn’t to do with placement on a device. And in any event, on Android, it’s extremely easy for users to move around where their apps are placed when they’re downloaded.

So again, we don’t think this gives rise to any questions of non-compliance with the DMA. our position is as I would have stated earlier.

As OWA writes in Google’s Hotseat Hypocrisy:

This is an overly narrow interpretation, and it’s particularly striking that Google chose to emphasise that the default browser is the one used when a user clicks on a URL, given that this is something Google itself actively undermines on Android. The Android Google Search widget, for instance, doesn’t open links in the user’s chosen default browser but instead opens them in the Google Chrome in-app browser.

Many other popular apps also override the system default browser and silently open links in their own in-app browsers, a practice we’ve documented extensively. As a result, the very concept of a default browser is being steadily eroded on mobile. In this environment, securing a position in the hotseat has become a crucial source of traffic for the user’s selected browser and a key part of respecting the user’s choice.

Artificial Intelligence

The second session of the day was on “Artificial Intelligence”. My distaste for mad billionaires stealing intellectual property, and then burning the globe to turn that stolen IP into bullshit is unbounded. But one question interested me because it’s about Search consuming and regurgitating websites, instead of sending traffic (and thereby starving them of the revenue to keep publishing).

Angela Mills-Wade from the European Publishers Council asked:

So notwithstanding Oliver’s claim that AI overviews are an evolution of search, they stand out as representing a kind of separate product and they compete directly in real time with primary source journalism and this is increasingly diverting audience attention away from publishers’ own channels.

This is not a DMA point, but it is important for the context. They’re representing information without any of the editorial standards or legal liability or accountability that define journalism. It’s also impacting Google’s search referral.They’re dipping (the referrals), sometimes as much as 50% to 60% and rising in some countries, which means both advertising revenues and opportunities for publishers to convert visitors to paying subscribers fall off a cliff.
The Google search engine seems to be accelerating the role of extractive products like AI overviews, and of course in the United States we also see AI mode, and they rely on content created by others to generate answers in their own ecosystem for Google to monetize.

There is no specific mechanism for publishers to opt out of all AI training while retaining inclusion and ranking in traditional search. So Google is both gatekeeper and competitor, self-preferencing, I know we’re not talking about that, AI product in competition with and obscuring primary source journalism.

And this unfair extraction and use of publisher content is without real consent, control or compensation to produce these new separate services which undermine not only copyright law but also the FRAND principles under the DMA, self-preferencing as I’ve mentioned, but also the lack of transparency.
…How does Google justify that full narrative answers AI overviews at the top of their search page means referrals to original journalism are falling fast, which reduces the economic return for publishers to be present on the search.

How does Google justify the extraction of publishers’ content in breach of copyright law and FRAND? And why does Google purchase third-party data for its own content services, for example finance, while treating publisher content as free input?

Why can’t publishers opt out of all AI if they would like to without losing their appearance in traditional search?
And will Google be respecting the publisher’s rights reservations, which is publishers right to do so under article 4 of the copyright directive or develop some granular controls specifically for AI overviews?

Olli from Google replied:

is there disruption to the news industry? Is there a need for an evolution within the ecosystem when it comes to the ways in which traffic goes to various parties and players? Yes, that has been the case for some time. We believe in journalism, the value of journalism. It’s a big part of Google’s success for there to be high-quality content on the Internet, but the internet is evolving and changing.

As part of that, we have done our best. I think we have made good steps, taken a creative and open-minded and have invested heavily, in fact, in ensuring that partners that we work with are able to evolve with those evolutionary ecosystem changes as they’re occurring around us, whether it’s news funds, new forms of partnership agreements, a raft of different initiatives that globally we’ve been very invested in to ensure that news publishers can innovate and evolve in this highly disruptive, technologically advancing environment.

Referral traffic itself and how those traffic flows have evolved over time. I will say this, and I’m happy to follow up, but the picture is somewhat more nuanced from our perspective. We see that AI overviews lead to a greater diversity of sites appearing in search.

We see that the quality of the traffic that is being sent to websites is of a higher quality. People tend to stay longer and we’re seeing more links in the web results so more opportunities for websites so when we’re thinking about referral traffics that of course not a single site analysis or even a subset of sites analysis you look in the round to see how the ecosystem has benefited from some of these innovations and i would posit that the picture is a little bit more nuanced than than perhaps is being represented here.

I’m not sure that really assuaged anyone’s worries. In fact, there are reports that an organisation called Independent Publishers Alliance and others have filed complaints to EU and UK about AI Overviews.

Data portability

The afternoon was devoted to data portability and sharing Google search index data, which wasn’t so interesting for me personally (it was a hot day, and I was tired after my friend’s wedding). The fourth session was a look at a new data portability tool, and its synthesised voice narrator was so weird I turned it off.

Initial impressions

I was disappointed that Google has chosen to view the hotseat issue so narrowly (and to continue having Chrome open links in its pre-installed Search app, even when it’s not default). This highly legalistic approach was demonstrated by the lack of any technical people in the Google team this year.

Apart from the predictable handwringing that every Gatekeeper felt obliged to perform about how EU consumers won’t receive the latest AI bells and whistles at the same time as Americans do, because of beastly regulation, Google made a good point about the regulatory regime:

My third point concerns the risk of fragmentation arising from enforcement of the DMA at the member state level. We see an increasing risk of regulatory fragmentation from national enforcement actions and private litigation. As the DMA allows for national level involvement, it clearly also designates the European Commission as the sole enforcer, with the goal of harmonising rules across the European Union. That is explicitly stated in Recital 6. This harmonisation objective is now being called into question as a result of an ever-increasing number of actions at the national level.

This matters for two critical reasons. First, the DMA’s core objective under Article 114 of the Treaty of the Function of the European Union is to prevent internal market fragmentation from divergent national rules. If national DMA litigation goes unchecked by the Commission, the central objective will fail.

Second, a potential explosion of national enforcement cases would create significant new regulatory overhead, directly contradicting the EU’s simplification package initiative. More practically, if a third party can simply take an issue to a member state court, potentially circumventing the Commission’s harmonising role, this undermines the incentive that gatekeepers have to engage constructively with the European Commission.

Publishing clear guidelines for national authorities and courts on engaging with the Commission on DMA compliance matters would send a very strong signal for coordination.

It seems to me that many of these national enforcement cases are due to frustration about how little has changed in the last 16 months. Microsoft is still self-preferencing its Edge browser on Windows; Google had showed encouraging signs, but its “letter of the law” approach is worrying; and Apple still doesn’t really allow other browser engines to ship on iThings, it still chokes Progressive Web Apps, and is litigating everything that moves in Brussels.

“Regulatory dialogue” is, of course, vital. But over a year after the DMA came into force, we need enforcement action by the Commission, as well as conversation.

Unofficial transcripts

I made these transcripts using MacWhisper from the official video recording. I manually corrected the passages I quoted above, so those are more accurate than these raw transcripts. Apologies if the software or I mis-spelled a name!

Session 1 (Choice screens)
Session 2 (Interoperaility)
Session 3 (data portability) Raw, uncorrected

Bruce Lawson's personal site