Google Must Open Android’s AI and Search Data to Rivals by 2027

The European Commission ordered Google on Thursday to open Android’s artificial intelligence features to rival assistants and share Google Search data with competitors. Both rules stem from proceedings the Commission opened in January, and they carry no fine of their own. Search data sharing begins in January 2027. Android’s AI gates open that July.

Brussels wants smaller AI firms and search engines to compete with Gemini and Google Search on merit instead of access. Google has had years to get there voluntarily, with little to show for it.

Two Orders, Different Clocks

The Commission adopted both decisions on July 16, eleven days ahead of its own statutory deadline. Both trace back to specification proceedings opened on January 27, giving regulators six months to write the technical detail gatekeeper law rarely spells out this precisely.

The Android decision breaks into four access points that currently separate Gemini from every other assistant on the operating system.

• Invocation – the wake word and home button long press that summon Gemini and nothing else
• Context – reading on screen content and app signals so a rival assistant can act on what a user is doing
• In-app actions – completing tasks inside other apps, such as booking a taxi or drafting a chat reply
• System resources – the on-device hardware and software access Google currently reserves for its own AI

Google must open all four to competing assistants by July 2027, under a decision covering eleven distinct Android features in total.

Why Brussels Chose Rules Over a Fine, This Time

Neither decision issued Thursday carries a monetary penalty. They are binding specification measures, spelling out exactly what Google must build rather than punishing what it already did.

Google still objects. In its own post on the ruling, the company argued that phone makers already safely vet AI assistants, and that handing external apps deep device permissions strips out protections manufacturers currently control.

That restraint has a backstory. Reports emerged months earlier that Commission President Ursula von der Leyen had personally intervened to stop a multi-billion euro penalty against Google that was already prepared, after investigators said they had gathered evidence Google was systematically breaching its gatekeeper duties.

A different fine looms regardless, and it could land within days. Sources told AFP the Commission is preparing a penalty in a separate case over search self-preferencing and Play Store rules, confirming Financial Times reporting from July 15. That penalty, expected in the hundreds of millions of euros, would cover Google favoring its own Shopping, Flights and Hotels listings over rivals, and blocking Play Store developers from steering users toward cheaper payment options. It would exceed the current DMA record of half a billion euros, set when regulators fined Apple over App Store steering rules in April 2025.

Thursday’s two orders carry no fine at all. But if Google refuses to comply with either one going forward, the ceiling is steeper than anything issued so far: up to 10% of Alphabet’s global annual turnover, or roughly $35 billion at current revenue.

Google has paid before. A decade of EU antitrust cases has cost the company roughly €11 billion, including a €4.1 billion Android bundling fine the bloc’s top court upheld just two weeks ago. Google has already made one DMA-driven concession this year too, opening its Play Store catalog to rival app stores while keeping its own commission structure intact.

The Voluntary Deal Google Already Tried

Google didn’t wait for a formal order to attempt compliance. It made a voluntary offer years ago, and according to the Commission’s own account of those talks, that offer meant removing between 90 and 100% of unique search queries from the shared dataset.

The same proposal excluded AI chatbots offering search functions from the pool of eligible recipients, narrowing who could even apply. Two years of talks produced no meaningful uptake from anyone the data was meant to help.

The new decision leaves less room for that outcome to repeat. It specifies its own pricing formula, its own anonymization method and an independent audit regime, conducted before access is granted and repeated annually for as long as a rival keeps using the data.

Who Gets a Seat at Google’s Table

The obvious winners are companies already building search-like AI without Google’s underlying data. OpenAI and Anthropic would have a stronger case for competing with Gemini on Android, while Microsoft could use fresh access to Google’s search data to sharpen Bing.

The stakes are large simply because of scale. Android runs an operating system used by 60% of the EU’s smartphone owners, so any rival assistant locked out of deep integration has been fighting for a fraction of a market Google’s own app reaches by default.

Gemini’s own reach already extends deep into that installed base, including Gemini’s expansion to 2GB budget Android phones, the kind of low-end hardware where a challenger would need equal footing to matter at all.

Fiona Scott Morton, a Yale economist and senior fellow at the Brussels based think tank Bruegel, published an analysis of the Android specification arguing the shift is fundamentally economic. When commercial success depends on product merit rather than who already controls distribution, the return from building an AI company in Europe rises, and so does the incentive to invest in one.

Will Europeans Get More Choice?

Not necessarily, and not soon. The rules take a year or more to bite, Google has already signaled it will fight rather than simply build, and Europe has already watched several AI products get pulled or delayed rather than opened under similar DMA pressure.

Kent Walker, Google’s president of global affairs, called the decisions a threat rather than a fix, telling Reuters:

Today’s decisions risk undermining vital privacy and security guardrails for millions of Europeans.

Walker added that Google had repeatedly offered solutions to safeguard users while meeting the law’s goals, and argued the rulings discount extensive evidence of user harm. Separately, Google says its own security researchers broke the Commission’s planned anonymization method in under two hours.

Outside experts split sharply on whether that risk is real.

• Google, Chamber of Progress and CCIA Europe – warn the access and data sharing rules create genuine privacy and security exposure, and that companies may pull products from Europe rather than open them
• DuckDuckGo – says the Commission’s anonymization approach already reduces reidentification risk to an insignificant level
• Knight-Georgetown Institute researchers – call the planned safeguards strong enough to support real competition, and note Google can independently verify the anonymization works as intended

Kay Jebelli, Europe vice president at the Chamber of Progress, a tech industry coalition that counts Google and Meta among its members, warned that the Commission is ignoring well documented privacy risks rather than working with industry on solutions the market actually wants.

Daniel Friedlaender, who heads the Brussels office of CCIA Europe (the Computer and Communications Industry Association), made a similar point on LinkedIn, arguing the Commission treats privacy and cybersecurity as secondary implementation questions rather than conditions for enforcement.

Dirk Auer, a director of competition policy at the International Center for Law and Economics, put the risk most bluntly. He said the likely result is fewer AI assistants for Europeans, not more, since companies facing compliance costs and legal uncertainty may simply withhold products instead of building them for the EU market.

Apple Chose a Different Path

Google’s strategy contrasts sharply with Apple’s. Apple tried to negotiate a compliant version of its new Siri AI before launch. Google launched Gemini integration first and left the compliance argument for later.

Apple’s approach produced a different outcome: no Siri AI in Europe at all. The company says regulators wouldn’t accept any proposal that preserved privacy and security, and its next software generation will simply skip the bloc.

Apple has gone further, publicly calling the Android AI interoperability measures a privacy and security disaster, even though the rules target its biggest rival rather than Apple itself.

The motive isn’t hard to guess. Brussels is pushing for similar openness on iOS next, and Apple’s business model depends on keeping its platform closed. Google’s own home market shows a milder version of the same staging pattern. The Android 17 rollout to Pixel devices staged its AI features for later, arriving on new hardware before the assistant layer caught up, even without EU intervention forcing the delay.

Google’s Countdown Starts Now

An appeal is widely expected, but it won’t pause the clock. Unlike traditional antitrust rulings, DMA obligations don’t automatically suspend when a gatekeeper challenges them in court.

A July 8 ruling from the EU’s General Court narrowed Google’s options further. In three cases brought by Apple, the court held that a gatekeeper cannot contest a DMA obligation in the abstract before the Commission issues a specific enforcement decision. Google must comply with both orders while any appeal proceeds before EU courts.

Google must finalize its pricing offer for the shared search data and send it to the Commission and prospective rivals within six months of adoption, the first deadline on a clock Brussels is now keeping instead of Google.

For now, Europe’s Android users keep Gemini as their only system-level assistant, and Google keeps its search data to itself. That changes, on paper, in six months.