For the past year, anyone with an Android phone in Europe has been living in a one-AI town. Google's Gemini is the only assistant that can really understand what's on your screen, control your apps, and wake up to your voice. That's about to change — by force.
On July 16, 2026, the European Commission dropped two legally binding specification decisions under the Digital Markets Act (DMA) that order Google to pry open Android's inner workings for rival AI assistants and share a decade's worth of search data with competitors. The deadlines are tight, the fines could run into tens of billions of euros, and the fallout will reshape how every European user interacts with their phone.
Under the first decision, Google must grant third-party AI assistants — think ChatGPT, Claude, Perplexity — access to 11 specific Android capabilities that were previously Gemini-exclusive. These break down into four buckets: invocation (hotword activation, Circle to Search), context (on-device app data, proactive suggestions), actions (cross-app tasks, background operations), and resources (access to on-device AI models and necessary hardware). In plain English, a future Android update will let you turn on your phone with a rival's wake word, have that assistant book a taxi through your ride-hailing app, or answer questions about a restaurant you just looked up — all without ever touching Google's software.
The timeline is aggressive. Most features must land in the next major Android release (presumably Android 18), no later than August 1, 2027. One narrow feature — concurrent always-on hotword detection — gets a one-year extension until 2028. But the broader interoperability package is due by mid-2027.
The second measure forces Google to share anonymized click, query, ranking, and view data with any compliant search engine or AI chatbot that offers web search. Pricing must follow FRAND (fair, reasonable, and non-discriminatory) terms — essentially incremental cost, capped at Google's own search operating margin for the largest beneficiaries, and fixed for five years. Small startups get even sweeter deals: strict incremental-cost pricing. This means a European AI search outfit like Mistral or a fast-growing challenger like Perplexity can tap into the world's richest search data trove to refine its retrieval systems, beginning January 2027.
[SPONSORED]
NEXT-GEN NPU CHIPSETS
Empower your local devices with desktop-class inference capabilities.
Google, unsurprisingly, is not thrilled. Kent Walker, President of Global Affairs, warned that the measures “risk undermining vital privacy and security guardrails for millions of Europeans” and that “Europeans' private searches would be exposed to unfamiliar companies, without adequate anonymisation.” The company's blog post argued that granting external apps sensitive device permissions threatens security, especially as the EU's own cybersecurity agency cautions about AI-era risks. Google even pointed a finger at OpenAI, claiming to The Register that the probe was “at least partly driven by OpenAI” attempting to use the DMA in unexpected ways.
Those concerns aren't theoretical. On Hacker News, one developer noted that “Google seems to be leveraging Play Integrity here… This is clearly anticompetitive.” Another thread, discussing a Chinese AI assistant that briefly gained system‑level injection rights, lit up with warnings: “If that permission opens to regular apps, your phone can be completely taken over — injecting bank passwords, tapping confirmations.” Enterprises are nervous, too. Roman Stanek, CEO of Good Data AI, told CSO Online that CISOs need to start treating “AI assistant” as a category risk, not a single trusted permission.
The contrast with Apple is stark. Apple simply refused to ship its new Siri AI features in the EU, citing DMA compliance burdens and privacy risks. Google, by contrast, got a year-long runway to implement changes while continuing to expand Gemini globally. As The Verge put it, “Google is better than Apple at playing the AI regulations game.”
The winners? European champions like Mistral, which has called for a “sovereign AI ecosystem,” and cross-border players like Perplexity, whose AI‑native browser and search product could slot directly into the new Android infrastructure. Phone makers like Samsung and Xiaomi, however, may face an awkward balancing act: promote their own in‑house AI assistants (Bixby, XiaoAI) or cede that real estate to third parties that now get equal billing.
Not everyone is celebrating. The think tank ECIPE argues the DMA could backfire, forcing Android to become more closed as Google hardens its security model in response to the mandate. And developers have noticed that Google's own AppFunctions API — the official bridge to let apps serve as tools for any assistant — remains in a “trusted tester” preview, limiting who can actually build for this new open future.
[SPONSORED]
▶ ENTERPRISE GPU CLUSTERS ◀
Scale your AI model training seamlessly. Book a Demo.
The next 12 months will be a scramble. Google may appeal, negotiate technical details, or play for time. Regulators in Brussels, fresh off a €4.1 billion fine against the company and probing self‑preferencing in search, don't look ready to blink. By the time Android 18 lands in pockets across Europe, the mobile AI assistant market will look nothing like it does today. The only question is whether it will feel like a breath of fresh air — or a security patchwork held together with duct tape.