The EU Just Made the Android Assistant Slot a Regulated Market — Two DMA Decisions, Explained

On July 16, 2026, the European Commission closed a set of proceedings it had opened against Google back in January by adopting two binding specification decisions under Article 6(7) of the Digital Markets Act. Article 6(7) is the provision that lets the Commission spell out, in operational detail, what a designated gatekeeper must actually do to comply with an obligation it already carries. These are not preliminary views and they are not fines. They are orders specifying concrete behavior, and non-compliance under the DMA can escalate to penalties of up to ten percent of global annual turnover, rising to twenty percent for repeat breaches.

The coverage was loud but imprecise, so it is worth stating the two decisions in the terms the Commission itself used.

Decision one: eleven Android features, opened to rival assistants

The first decision requires Google to give third-party AI assistants effective access to eleven Android operating-system features that today its own Gemini assistant uses fully and rivals cannot. The Commission frames this in terms of what an assistant needs to be a real assistant rather than a chat window: the ability to hear a user, to understand context, and to act on the user's behalf.

The eleven features span voice invocation, so a rival can be triggered hands-free; App Functions and cross-app action, so an assistant can perform a task inside another app; access to contextual data, so it knows what the user was just doing; background execution; and access to device resources. The Commission offered concrete examples of the behavior it wants unlocked: booking a taxi by voice, receiving suggested replies inside a chat app, and asking about a place the user recently visited. The measures are grounded in the roughly sixty percent of EU smartphone users who carry an Android device.

Crucially, five of the eleven features come with a safeguard: for the most sensitive capabilities — including screen automation, structured integration, system integration, centralized access to app data, and context-aware intelligence — Google may set objective and non-discriminatory privacy, security, and integrity criteria governing which assistants qualify. Covered access also requires explicit user consent. That safeguard is reasonable on its face, and it is also the seam where the next two years of enforcement disputes will run: the whole fight will be about whether Google's criteria are genuinely objective or a slower-motion version of the same exclusion.

Decision two: Search data sharing, from January 2027

The second decision, which drew less coverage but is arguably the heavier of the two, requires Google to share with eligible third parties — explicitly including OpenAI and other AI chatbots that offer search functionality — the same category of data it collects to optimize its own Search. Sharing begins in January 2027, subject to a multi-layered anonymization method and a fair, transparent pricing formula set out in the decision.

This matters because query-behavior data is the one input a challenger cannot manufacture at any price. A search-capable assistant is only as good as what it can retrieve, and Google's real advantage in retrieval is not its ranking code but the accumulated signal of billions of queries. Ordering that data licensed to rivals reaches past the interface and into the foundation of Google's position. The pricing formula and anonymization method are, predictably, the most contested lines in the entire order.

The timeline the headlines got wrong

The single most-repeated claim — that Europeans will soon say "Hey ChatGPT" instead of "Hey Google" — attaches to the wrong year. The bulk of the Android interoperability work is tied to Android 18 and is due no later than August 1, 2027. But concurrent wake-word access — the specific capability that lets separate assistants respond to different wake words at the same time, which is what makes "Hey ChatGPT" work as a live alternative — carries a later deadline of August 1, 2028, tied to Android 19. The headline feature is the last one to arrive, a full year behind the rest of the package.

The staged rollout is not a detail. A remedy that phases in over two years lets the incumbent bank its habit advantage during the gap. Between the bulk of interoperability landing in August 2027 and concurrent wake words landing in August 2028, a rival assistant on Android can act inside apps and read context but still cannot own a hands-free trigger of its own. For that year, the most reflexive surface — spoken invocation — stays effectively Google's.

Google's response

Google objected the same afternoon, through Kent Walker, its president of global affairs, on privacy and security grounds rather than competition. The argument is that an AI assistant with system-level access to the microphone, screen, and app data is among the most sensitive software that can run on a phone, that phone makers already vet which assistants get deep access, and that detaching sensitive permissions from that vetting risks user safety. It is a legitimate tension — the hooks that let a good assistant help are the same hooks a bad one could abuse — and it is also the argument a moat-holder reaches for when the moat itself is the thing being ordered open. The Commission's answer was the explicit-consent requirement and the objective-criteria carve-out on the five sensitive features.

Why this is different from the old Google fines

The EU has fined Google before — €2.42 billion over Shopping in 2017, €4.34 billion over Android in 2018, €1.49 billion over AdSense in 2019 — and the common weakness of those cases was that they punished past conduct without reliably changing future behavior. Google paid and, in the ways that mattered, continued.

The DMA specification model is built to fix exactly that. It does not wait for a court to find abuse after the fact; it specifies the required behavior in advance and escalates if the behavior does not appear, against a penalty ceiling measured as a share of a roughly four-hundred-billion-dollar revenue base. That is what distinguishes this from a cost-of-doing-business fine. It is also why the more important story is not the money at all but the target: this is the first major regulatory action aimed not at what an AI model can do, but at who gets to be the assistant a user reaches for by default.

What to watch

Four checkpoints will tell you whether this becomes a turning point or a regional footnote. First, in January 2027, whether Search optimization data actually begins flowing to OpenAI and others, and whether the pricing formula is one a rival will pay. Second, in mid-2027, how objective Google's criteria for the five sensitive capabilities turn out to be when published. Third, at the August 1, 2027 deadline, whether voice invocation and cross-app action actually work for a rival assistant on an ordinary European Android phone, or whether compliance is technically present and practically hollow. Fourth, at the August 1, 2028 deadline, whether concurrent wake words arrive — and whether Google confines the capability to Europe or ships it globally. That last answer decides whether the assistant slot becomes a contestable market worldwide or stays a Google default everywhere the DMA does not reach.

For the strategic read on why the assistant slot, and not the model, is the asset worth regulating, see the companion analysis: The Wake Word Is the Moat.