EU Pressures Google To Open Android’s AI To Rivals, Google Calls It “Unwarranted”

The European Commission wants Google to loosen Gemini’s grip on Android in Europe, arguing that the company’s AI assistant enjoys system-level privileges competitors cannot match. Google is pushing back, calling the proposed changes an “unwarranted intervention.”

The move is part of an ongoing investigation under the EU’s Digital Markets Act (DMA), the bloc’s flagship competition framework for large tech platforms it labels “gatekeepers.” Following an initial probe that began in January, regulators have outlined how they believe Android should open up to third-party AI services from voice triggers to hardware access.

Today, when a user switches on a Google-powered Android phone, Gemini is already present and plugged deeply into the system. The European Commission sees that as a structural advantage for Google’s own AI, going beyond preinstallation to include exclusive capabilities.

According to the Commission, a range of Android experiences currently work only with Gemini. While users can install other AI apps like ChatGPT or Grok, those services do not get the same access to data, system integrations, or controls. Regulators highlight use cases such as sending an email in the default mail app or sharing a photo with friends, where Gemini is effectively the only integrated route.

Henna Virkkunen, the Commission’s vice president for Tech Sovereignty, framed the effort around interoperability and user choice. She said that as AI evolves, interoperability is “key” to unlocking its potential, adding that the proposed measures are intended to open Android devices to a wider range of AI services so users can choose tools that match their needs and values “without sacrificing functionality.”

The Commission’s stance builds on earlier DMA enforcement. Google has already been required in Europe to offer search engine choice screens on Android, support alternative payment methods in the Play Store, and limit certain data sharing across its services. AI integration on Android is the latest front.

The Commission has laid out a broad framework of potential obligations aimed at putting third-party AI tools on a more equal footing with Gemini. These proposals include:

• System-wide invocation of third-party AI: Regulators want users to be able to trigger non-Google AI assistants at the system level, using hot words or hardware buttons, not just by opening an app. That would bring rivals closer to the integration level Gemini currently enjoys.
• Access to on-screen context: When a user invokes an AI assistant, the system could allow it to view relevant on-screen content, helping it respond in context. This is similar to contextual assistance that Google already provides via Gemini-powered features.
• Deeper access to local data for proactive features: The Commission suggests allowing alternative AI systems to access local data (under user control) to offer proactive suggestions and summaries. The description in the report resembles Google’s own “Magic Cue” feature, which relies on Gemini to propose actions based on current activity.
• Control over apps and system features: Google has started testing capabilities where Gemini can operate apps on a user’s behalf, such as controlling certain installed applications. While early tests, including on Samsung’s Galaxy S26, show Gemini still struggles with this reliably, the Commission wants to explore enabling other AI providers to autonomously control apps and system functions as well.
• Hardware access for local AI models: Many Gemini features on Android run on-device, using local models. Regulators note that Google has been slow to give others the system access needed to run their own local models effectively. They are therefore considering a mandate that would guarantee developers access to the necessary hardware resources so their models can run with “high levels of performance, availability and responsiveness.”
• New APIs and free technical assistance: Under the DMA, Google may have to build new Android APIs to let rival AI services plug into the OS, and provide technical assistance to support that integration. The Commission specifies that these tools and support should be made available free of charge.

In essence, Brussels is seeking to turn some of Gemini’s privileged hooks into platform capabilities that any qualifying AI assistant can use, subject to user choice and consent.

Google’s response centres on how these obligations might affect device makers and users. Claire Kelly, Google’s senior competition counsel, argued that the proposed measures would undermine Android’s flexibility for OEMs and for Google itself.

She said the intervention would “strip away” device makers’ autonomy to customize AI services, and would mandate access to “sensitive hardware and device permissions.” According to Google’s position, this could “unnecessarily” increase costs and weaken “critical privacy and security protections for European users.”

Google also points out that users can already install other AI chatbots on Android and choose to use them. The core disagreement lies not in installability, but in whether rivals should gain deeper, system-level access similar to Gemini, and under what safeguards.

The Commission’s Android AI plan is not finalized. The current document is a framework for potential obligations, and regulators are accepting feedback from interested parties until May 13. A final decision on this particular DMA investigation must be made by July 27.

If the Commission confirms that Google must implement the proposed changes, non-compliance could carry significant penalties. The DMA allows for fines of up to 10 percent of a company’s annual global revenue for violations, though no specific fine has been announced or proposed in this case.

Even with a final decision, any technical overhaul is likely to be gradual. Giving third-party AI services new access to system tools, local data and hardware accelerators requires careful engineering and testing. Moving too fast could introduce security or privacy issues, something both regulators and Google acknowledge as a risk. There is also no assurance that such changes, if adopted, would extend beyond the EU; they could remain region-specific.

For now, Gemini keeps its privileged status on Android in Europe, but that could shift as the DMA’s AI test case moves toward a summer decision.