The European Commission has launched action under the Digital Markets Act aimed at forcing Google to open up parts of Android's features to rival AI assistants. The stakes: around 60% of European smartphones run on Android, and deep integration of Gemini could lock down that access. Google disputes this, citing security and privacy risks.
There is a considerable difference between offering the best AI assistant and being the one that's already installed. That's exactly what this case is about, and it directly affects what you'll be able to do with your phone in two years.
The framework: what is the DMA
The Digital Markets Act (DMA) is a European law that came into force in 2023. Its principle is simple: certain very large platforms, designated as gatekeepers, hold a position that lets them lock down a market. The regulation therefore imposes opening obligations on them that other companies don't have.
In practice, this means a company like Google can't just invoke commercial freedom: beyond a certain size, it must allow competitors access to features of its own platform. This is a different logic from classic competition law, which punishes abuse after the fact. Here, openness is imposed upstream.
What is being challenged
The action concerns access to Android features for third-party AI assistants. The Commission's reasoning is as follows: an assistant integrated into the operating system has advantages that a competitor installed as a mere app cannot match. It can be invoked with a native gesture, access system data, and tie into the phone's other services.
If Gemini benefits from this deep integration while competitors have to make do with surface-level access, the competition isn't fair, regardless of model quality. With around 60% market share on European smartphones, Android's scale turns a technical advantage into a dominant position in an entirely new market.
The case is said to involve around ten Android features as well as data-sharing rules, with compliance timelines.
Google disputes this, citing security and privacy arguments. And they're not absurd. Opening system features to third-party apps mechanically widens the attack surface. An assistant that accesses your notifications, messages, and system data is a sensitive entry point, and not all developers offer the same guarantees. This is the permanent tension of platform regulation: openness favours competition, closure makes security control easier. Neither position is purely cynical nor purely virtuous.
Why this case matters more than it seems
What's at stake here is the access layer to AI. We're entering a period where the assistant becomes the primary interface between you and your devices, as we noted regarding smart glasses, where the brand already in your pocket starts with a decisive lead.
And positions crystallise quickly. Google has signed a deal for a Gemini model to power Apple's future Siri, as we discussed in our article on Gemini's delays. If Gemini powers both the Android ecosystem and Apple's assistant, the vast majority of smartphone users worldwide would interact with a model from the same provider. That's not illegal in itself, but you can see why a regulator would be concerned.
The European question, in both directions
This case illustrates the European position, with its strengths and limits.
On the strengths side, Europe is currently the only jurisdiction imposing structural opening obligations on dominant platforms. That creates room for alternative players, including European ones, that wouldn't otherwise exist. An AI assistant developer has a better chance of reaching a European user than an American one, precisely because of this framework.
On the limits side, regulating isn't building. We noted in our article on Genesis-Science-1 that the US is now mobilising its public labs to produce open models, and that China is making openness an industrial strategy. Guaranteeing access to a market is useful; you still need something to offer in it. Europe has Mistral and a few solid players, but the gap in resources with the American and Chinese giants remains considerable.
What to take away
For you, the user, the issue is very concrete: in three years, will you be able to freely choose which AI assistant runs your phone, or will you be limited to the one the manufacturer has integrated? That's the kind of question that seems secondary until it's settled, and becomes irreversible once it is.
Both sides have credible arguments, and the decision that comes out of this case will likely set precedent well beyond Europe. What's certain is that the issue isn't limited to a technical quarrel between a commission and a company. It's about who gets the right to talk to your phone, and therefore who gets to see your questions.