The European Commission has published its preliminary conclusions on how Google manages artificial intelligence on Android. According to the organization, the operating system favors Gemini over the rest of its competitors, which is why it requires the company to apply measures to promote interoperability between other AI alternatives in its ecosystem. As might be expected, Google, for its part, is not willing to accept it without a fight.
Another chapter in the Digital Markets Law. This law (DMA) is what forces large technology companies considered “gatekeepers” (including Alphabet) to guarantee fair conditions of competition on their platforms. Google has been subject to this legislation since March 2024 and because of this it has had to introduce changes in Europe, such as showing screens so that the user can choose other search engines apart from Google on Android, or allowing alternative payment methods in its application store.
Now Europe has knocked on the door again, this time over questions about Google’s AI, and it is the next chapter in this tug-of-war between regulation and private companies.
Gemini rules Android. When you turn on an Android mobile with Google services, Gemini is already there, integrated at the system level. It can be voice activated, access screen context, interact with other apps, and generate proactive suggestions based on your activity. Apps like ChatGPT or Claude’s assistant can be installed, but they do not have the same level of access.
The European Commission points out specific cases where Gemini is the only available way: sending an email from the default email app, ordering food at home or sharing a photo with contacts. That, according to Brussels, is not fair competition.
What the EU proposes. The preliminary measures published last Monday point in several directions. Third-party AI services should be able to be activated using custom wake words or physical buttons on the device. They should also be able to access screen context when the user opens them, and query local device data to provide suggestions and summaries, something only Gemini now does.
In addition, the Commission proposes that other AIs can control apps autonomously, as Gemini is already beginning to do (although the result still leaves something to be desired in some cases) and that external developers have access to the hardware necessary to run local models with comparable performance.
Finally, Google could be forced to create new APIs and provide technical support to other AI developers who want to integrate into Android, all at no cost to third parties.
Google’s response. The company was quick to react. Clare Kelly, senior competition adviser, described the proposal as an “unjustified intrusion” that would “force the granting of access to sensitive hardware and device permissions, unnecessarily increasing costs and undermining critical privacy and security protections for European users.” Google defends that Android is already an open ecosystem and that device manufacturers have full autonomy to customize the AI services they offer to their users.
What’s coming now. The process is not over. The Commission is opening a public consultation until May 13, after which it will review the input they have received (including from Google) before issuing a decision by July 27. If Google does not comply with the measures or an agreement is not reached, the company is exposed to fines of up to 10% of its global annual turnover.
As they share from Ars Technica, although Google would not have to open its systems suddenly, implementing these changes would take time and doing so in a hurry could create security risks. Furthermore, as is often the case with DMA decisions, any changes that finally occur would be, at least in principle, limited to the European market.
Cover image | José García and François Genon
In | A developer went to sleep with a $10 alert on Google Cloud: he woke up to a bill of more than $18,000
