EU Forces Google to Open Android to AI Rivals in Landmark DMA Decision

The European Commission fired its most consequential shot in the AI era on July 16, issuing two binding specification measures to Google under the Digital Markets Act that could fundamentally reshape how artificial intelligence reaches consumers across Europe. The rulings — covering Android AI interoperability and Google Search data sharing — give rivals including Anthropic and OpenAI a regulatory path to compete on the world’s dominant mobile operating system for the first time.

The Android Decision: Eleven Features Pried Open

At the heart of the first specification measure is a deceptively simple premise: AI assistants other than Google’s own Gemini cannot currently do what Gemini can on an Android phone. They cannot be activated by voice, cannot book taxis on a user’s behalf, cannot read and reply to messages, cannot perform the system-level tasks that make an AI assistant actually useful rather than merely present.

The Commission found that Google’s preloading of Gemini onto Android devices — reaching roughly 60 percent of EU adults who use Android smartphones — had materially reduced the attractiveness of rival third-party AI models. By controlling the operating system and the AI assistant simultaneously, Google had effectively converted Android into a distribution moat.

Under the new ruling, Google must open 11 Android features to competing AI services. These include the ability to activate via a wake word (the “Hey Google” equivalent), to access device sensors and microphones for real-time tasks, and to execute actions across third-party apps such as booking rides, scheduling calendar events, and composing messages through voice commands.

The timeline is deliberately structured: the full feature set must be available by the time Android 18 ships, targeting August 1, 2027. Concurrent wake-word support — allowing users to have multiple AI assistants active simultaneously without one silencing the other — carries a slightly later deadline of August 2028.

The Search Data Decision: Sharing the Crown Jewels

The second ruling addresses something that no AI company outside Google has ever had: access to the behavioral signals that make Google Search so hard to dislodge. Google must now share anonymized ranking data, query logs, click data, and view data with eligible rival search engines and AI chatbots that offer search functionality.

The data package, which Google uses internally to continuously improve its search algorithms, must begin flowing to competitors in January 2027. The Commission developed the anonymization methodology in close collaboration with privacy experts and in consultation with the European Data Protection Board to ensure GDPR compliance.

The implications are significant. For AI companies building retrieval-augmented generation products, web search quality is a critical differentiator. Access to Google’s behavioral data — even in anonymized form — could meaningfully narrow the gap that has kept challengers from unseating Google in web search for two decades.

Google Pushes Back

Google’s response was swift and sharp. Kent Walker, the company’s President of Global Affairs, said in a July 16 statement that “today’s decisions risk undermining vital privacy and security guardrails for millions of Europeans.” The company argued that opening system-level Android access to third-party AI assistants creates new vectors for malicious apps to impersonate trusted services and intercept sensitive data.

Those are legitimate technical concerns — granting deep OS access to any software raises genuine security tradeoffs — but antitrust regulators have heard similar arguments from Google before, most notably during years of litigation over Android bundling in Europe. The Commission has indicated it worked with external privacy experts to structure both decisions to avoid GDPR conflicts, a preemptive answer to Google’s objections.

Whether Google will comply, appeal, or attempt to slow-walk implementation will be worth watching. The DMA carries fines of up to 10 percent of global annual turnover for non-compliance, with the possibility of structural remedies for repeat violations.

Who Benefits — and By How Much?

The direct beneficiaries named in most analyses are Anthropic and OpenAI, whose Claude and ChatGPT products are the most widely used AI assistants outside the Google ecosystem. Under the new rules, a user could theoretically set Claude as their default Android AI assistant and have it perform the same system-level tasks that Gemini currently monopolizes.

But the practical gains depend heavily on implementation quality and consumer adoption. Even if the technical interoperability is perfect by August 2027, Google retains enormous distribution advantages: Gemini is the default, Android interfaces are designed around Google services, and the company can still prominently promote its own AI across the ecosystem in ways that don’t technically violate the specification measures.

The search data decision may prove more impactful in the near term. AI companies building products that depend on web search — a growing category as language models are increasingly used as research and information tools — could see immediate improvements in retrieval quality once Google’s behavioral data becomes available in January 2027.

A Template for AI Regulation

The July 16 decisions represent the first time the Digital Markets Act has been applied specifically to the AI assistant market. Prior DMA enforcement actions against Google addressed search self-preferencing and shopping comparison services — domains where the competitive dynamics were well-established. The AI market is newer, faster-moving, and far more technically complex.

By issuing specification measures rather than formal non-compliance findings, the Commission chose a forward-looking approach: defining what compliance looks like before issuing penalties. It is a model that other regulators, including the UK’s Competition and Markets Authority and the US Federal Trade Commission, will be watching closely.

The broader signal is clear. Across Europe, regulators now view large language model distribution as a competition problem as much as a safety problem. As AI assistants become the primary interface through which billions of people access information and accomplish daily tasks, the question of which assistant runs by default on which device has graduated from a product decision to a matter of public policy.

For Google, the Android AI ruling is not just another regulatory headache. It is the first serious regulatory threat to the distribution advantage that has made Gemini the de facto AI assistant for most of the world’s smartphone users — and it arrives at precisely the moment when AI assistant market share is becoming the most contested battleground in technology.