The European Commission issued two sets of binding specification measures to Google on July 16, 2026, under the Digital Markets Act (DMA). The decisions target something that has quietly tilted the playing field for AI adoption across hundreds of millions of Android devices: Gemini gets system-level access that every other AI assistant does not.

That is about to change — and if you are thinking about which AI tools to build your business on, the ruling matters more than it looks.

What the EU Actually Decided

The Commission’s two measures cover distinct but related problems:

Android AI interoperability. On current Android devices, Google’s own Gemini assistant plugs directly into operating system features: reading on-screen content, controlling apps, accessing device context, and responding to voice triggers in ways that feel native. Competing assistants — Anthropic’s Claude, OpenAI’s ChatGPT, Perplexity — have to operate as ordinary apps, with none of those deep hooks. The new measures require Google to extend the same system-level access to rivals. The changes are expected to reach users by July 2027.

Google Search data sharing. Google’s search engine generates an enormous stream of anonymised signals — queries entered, results clicked, rankings applied, content viewed — that it uses to continuously improve its own products. No competitor has access to this data at anything close to the same scale. From January 2027, Google must share those anonymised query and engagement signals with eligible third-party search and AI providers, under a regulated pricing formula.

Both measures were developed through a consultation process that included privacy experts and the European Data Protection Board, with anonymisation safeguards built into the data-sharing framework.

Why This Has Been a Real Problem

Google’s gatekeeper status on Android is not abstract. Roughly 72% of smartphones globally run Android. When a business deploys an AI assistant to its workforce — for internal knowledge search, customer-facing support, or operational workflows — the practical utility of that assistant often depends on how well it integrates with the devices employees actually carry.

An AI assistant that can read a document open on screen, pull context from a calendar, or hear a voice command without requiring the user to switch apps is meaningfully more useful than one locked inside a single app container. Gemini has had that capability by design. Its rivals have had to work around it.

For businesses evaluating AI tools, this created a subtle but real lock-in effect: even if a company preferred Claude or ChatGPT for quality or pricing reasons, Gemini’s Android advantage could push the decision toward Google’s stack.

What Changes for Business AI Adoption

The practical effects of these decisions will take time to land — July 2027 for the Android changes, January 2027 for search data — but the direction is now set by enforceable legal obligations, not voluntary commitments.

For businesses already deploying AI agents and assistants, a few things shift:

Mobile AI becomes more competitive. The assumption that “Gemini works better on Android” will weaken as rivals gain system-level access. Businesses choosing AI tools for field teams, customer-facing staff, or mobile-first workflows will have more genuine choice.

Search intelligence democratises. AI products that compete with Google Search — or use search signals to improve their models — gain access to data they have never had before. That affects everything from enterprise search to AI assistants that surface information from the web. Companies building AI-powered research, compliance monitoring, or market intelligence tools will eventually benefit from a more level data landscape.

Competitive pressure on Google’s AI suite accelerates. Google’s response to these measures will matter. If it has to genuinely open Android to rivals, the company will need to win enterprise AI customers on model quality, pricing, and integration — not platform control.

The Bigger Pattern

The DMA is the EU’s clearest statement that large platforms cannot use dominance in one market (mobile operating systems) to entrench dominance in adjacent ones (AI assistants). It follows similar pressures applied to Apple, Meta, and others in areas ranging from messaging interoperability to app store access.

What is new in 2026 is that AI has become central to the enforcement picture. The Commission’s language is explicit: rival AI services must be able to compete on “equal footing” with a gatekeeper’s own AI products. That framing suggests further action is likely as AI becomes more embedded in operating systems, browsers, and cloud platforms.

What This Means for Business

If you are building or buying AI tools for your business, the DMA decision is relevant in two ways.

First, it signals that vendor lock-in through platform control is increasingly a regulatory target. Choosing AI tools based on genuine capabilities — rather than which company controls your phone’s operating system — becomes a more defensible strategy.

Second, it confirms that the competitive AI market is being actively protected. More providers competing on equal terms means more pressure on pricing, quality, and specialisation. Businesses that invest in AI capability now — rather than waiting for a clear market winner — are building skills that transfer regardless of which model or assistant leads in three years.

The search data provisions are harder to predict in their effects, but they matter for any business that depends on AI-powered research, competitive intelligence, or content tools that rely on understanding search intent at scale.

---

The EU’s DMA enforcement is moving steadily from intent to obligation. The July 2026 measures on Google are the most direct intervention yet into how AI assistants compete on consumer devices. Whether you operate inside or outside Europe, the ripple effects on the AI tool market will be global.

If you are mapping your business’s AI strategy and want to think through what vendor independence actually means in practice, talk to Enterprise DNA’s advisory team about how to evaluate AI tools on capability rather than platform advantage.