Skip to main content

What the EU's New Google Ruling Means for Your Android Phone (and Your Searches)

Ilustrační obrázek
Imagine waking up to a phone that lets you choose which AI assistant helps you book a taxi, translate a conversation, or reply to a message — not because Google generously allowed it, but because Brussels made it a legal requirement. That is the reality the European Commission set in motion last week, and it will start reshaping what your Android phone can do from mid-2027.

Two decisions, one goal: more choice on your phone

On 16 July 2026, the European Commission adopted two binding decisions under the Digital Markets Act (DMA) that directly affect anyone in Europe who uses an Android phone or searches the web. The first forces Google to open core parts of its Android operating system to competing AI assistants — not just Gemini. The second requires Google to share anonymised search data with rival search engines and AI chatbots, something it has largely avoided until now.

The decisions come after the Commission concluded that Google's own voluntary compliance proposals were, in its words, ineffective. In one case, the search-data offer Google put forward stripped out between 90 and 100 percent of unique search queries, according to a detailed report by PPC Land. No meaningful uptake followed.

These are what the law calls "specification proceedings" — not fines, but detailed technical instructions that spell out exactly what a gatekeeper must do. The Commission is not accusing Google of breaking the law (yet); it is telling Google, in granular terms, what compliance looks like. Fines for non-compliance could follow later if Google drags its feet.

What actually changes on your Android phone

For the roughly 60 percent of European mobile users who carry an Android device, the practical change is this: rival AI assistants will gain access to the same operating-system features that Gemini enjoys today. The Commission identified 11 specific Android features across four categories that must open up:

Invocation — how you start an interaction. You could say "Hey Assistant" to a non-Google AI service and it would wake up, just like Gemini does now. By 2028, several assistants could even listen simultaneously for their own wake words — so you would not have to pick one in advance.

Context awareness — what the assistant knows about your situation. A rival AI could read data from your calendar, messages, or the screen you are looking at to give relevant suggestions. Imagine your assistant spotting a flight number in an email and offering to check its status, without you typing a thing.

Actions — what the assistant can do on your behalf. Booking a restaurant through an app, ordering a taxi, translating a live conversation — multi-step tasks that today only Gemini can handle smoothly would become available to competitors.

Access to resources — hardware and software power. This includes Google's own on-device AI models that run directly on the phone chip, not in the cloud.

Google must implement the bulk of these changes in Android 18 by 1 August 2027. The more complex concurrent-hotword feature has a longer runway — Android 19 by 1 August 2028. Users should start noticing the difference from July 2027, the Commission estimates.

"We tested" — what this means from a privacy angle

Google's response was swift. Kent Walker, President of Global Affairs at Google and Alphabet, warned the same day that the decisions "risk undermining vital privacy and security guardrails for millions of Europeans." He argued that Android already lets AI assistants access its capabilities safely through manufacturer vetting, and that the new obligation "threatens device security by granting external apps sensitive and powerful device permissions without these safeguards."

The Commission has anticipated this concern. For a handful of especially sensitive features — screen automation, structured on-device integration, centralised device data access — Google may impose objective, non-discriminatory eligibility conditions verified by independent certifiers. And crucially, you remain in control: users must grant explicit consent before any assistant can use any given feature. If you do not want an assistant poking through your messages, you do not have to allow it.

Device manufacturers also keep the right to customise Android's interface and pre-installed software, as long as those customisations do not block the required interoperability itself. Samsung, Xiaomi, and others cannot be forced to strip their own assistants from their phones.

The search data side: why it matters and who qualifies

The second decision targets a quieter but equally powerful monopoly. Google Search holds above 90 percent market share in the EU, and has for decades. That dominance generates a firehose of query, click, ranking, and view data that no competitor can replicate — and which AI chatbots increasingly need for factual grounding.

Starting January 2027, eligible rival search engines and AI chatbots with search functions can request that anonymised data from Google. The dataset covers what people search for, what results they see, where those results ranked, and how they interacted with them — across both free and paid search results. Several categories are explicitly excluded: no account information, no precise timestamps, no rare or long-tail queries that could identify individuals, no exact location data, and no URLs tied to paid results.

The eligibility bar is real. Applicants must run an online search engine in the EU for at least two consecutive years — or be a startup less than two years old that has raised more than €50 million in capital and attracted at least 50,000 monthly average EU users in the past year. Companies under international sanctions or controlled by countries deemed cybersecurity risks are excluded.

Pricing follows cost recovery plus a capped margin that does not apply to small and medium businesses at all. In plain language: smaller European players — think Ecosia, Qwant, Startpage — would pay near cost for data that could help them build competitive search and AI products.

What this means for you, practically

If you are an Android user in Europe, the most visible change will arrive around summer 2027, when you may see a wider choice of AI assistants pre-installed or downloadable on your phone. Some may be from companies you have never heard of — European startups, privacy-focused developers, or productivity tools that do one thing really well rather than everything.

If you value privacy, the safeguards are meaningful but not bulletproof. The Commission's anonymisation approach — a multi-layered technical method aligned with draft joint DMA-GDPR guidelines — will be audited independently within six months of any recipient starting to process the data, and then annually. It will also be reviewed every two years, and the Commission can reopen the file if independent testing shows the safeguards are not holding up.

If you are a European business building AI tools, the search-data provisions may be the bigger deal. For the first time, a startup could train its own search relevance or AI-grounded retrieval system on the same kind of data that powers Google's dominance — without needing to be Google. Whether that produces a genuinely more competitive market depends on execution, but the legal framework is now in place.

The bigger picture: Europe draws a line

The specification decisions do not exist in a vacuum. They arrive weeks after the Court of Justice of the European Union confirmed a €4.125 billion fine against Google over historical Android practices (a separate, unrelated case), and a month after the Commission used a different legal tool to order Meta to reopen WhatsApp to rival AI assistants. Brussels is signalling, across multiple cases and legal instruments, that access to users through dominant platforms — whether an operating system, a messaging app, or a search box — is a competition problem that requires structural remedies, not just fines.

The timeline matters too. Google must submit its search-data eligibility form by end of August 2026, provide licence templates and test samples by September, finalise the anonymised dataset by November, and begin sharing data by January 2027. On the Android side, draft certification terms for sensitive features are due by February 2027. This is not a distant five-year roadmap; it is a series of near-term deadlines with measurable outputs.

For the rest of us — the people who just want our phones to work well and our searches to respect our privacy — the decisions are a reminder that digital choice is not a natural market outcome. It is something regulators have to fight for, one specification at a time.

X

Don't miss out!

Subscribe for the latest news and updates.