The European Commission is preparing to tell Google exactly how it must open Android to rival AI assistants, escalating a regulatory confrontation that will determine whether artificial intelligence becomes the next great platform lock-in or the first to be broken before it sets. EU watchdogs are poised to lay out what Alphabet must do to grant the likes of OpenAI’s ChatGPT and Anthropic’s Claude access to the same Android features that Google reserves for Gemini, including voice activation, system-level search integration, and the ability to interoperate with other Android software, Bloomberg reported on Wednesday. The draft findings are part of specification proceedings opened under the Digital Markets Act in January, and they arrive at the precise moment Google is completing Gemini’s takeover of the Android assistant experience for more than two billion devices worldwide.

Google has said it is concerned the measures could “compromise user privacy, security, and innovation.” The Commission’s position is that a company controlling roughly 65% of Europe’s mobile operating system market cannot be the sole arbiter of which AI gets to talk to the phone.

Two proceedings, one deadline

The Commission opened two parallel specification proceedings on 27 January 2026, each targeting a different obligation under the DMA. The first, under Article 6(7), concerns interoperability: Google must provide third-party AI developers with “free and effective interoperability” to the Android hardware and software features that Gemini uses. The second, under Article 6(11), concerns data: Google must share anonymised search ranking, query, click, and view data with rival search engines and, critically, with AI chatbot providers on fair, reasonable, and non-discriminatory terms.

On 16 April, the Commission told Google what it must do to share search data with rivals, publishing preliminary findings in a 29-page specification document that defines at the field level what data must flow, how it must be anonymised, how it may be priced, and what auditing regime will govern it. A public consultation on those measures runs until 1 May. The Android AI interoperability proceedings, tracked separately as case DMA.100220, are following a parallel timeline. Bloomberg’s report suggests the Commission’s draft findings on that track are imminent. The final binding decision on both must be adopted by 27 July 2026.

Henna Virkkunen, the EU’s Executive Vice-President for Tech Sovereignty, Security and Democracy, announced the proceedings in January. Teresa Ribera, the Commission’s competition chief, framed the rationale plainly: “We want to maximise the potential and the benefits of this profound technological shift by making sure the playing field is open and fair, not tilted in favour of the largest few.”

What “equally effective access” means in practice

The interoperability question is the more consequential of the two proceedings. Search data sharing, while commercially significant, is a question of inputs. Android interoperability is a question of position. Today, a user who downloads ChatGPT or Claude on an Android phone gets an app. A user who uses Gemini gets an operating system feature. Gemini can be invoked by holding the power button or saying “Hey Google.” It can read the screen, interact with other apps, and access system-level functions that third-party assistants cannot. The DMA’s Article 6(7) says that asymmetry is not allowed if Google is using it to favour its own services.

The Commission intends to specify how Google must grant rival AI providers equally effective access to those same capabilities. That could mean letting users set ChatGPT or Claude as the default system assistant, giving third-party AI services the same hooks into voice activation and always-on listening, and allowing rivals to integrate with Gmail, Calendar, and other Google apps in the way Gemini does natively. It is the difference between being an app in a drawer and being the intelligence layer of the phone.

Google argues that “Android is open by design” and points to the fact that users can already download any AI app from the Play Store. The Commission’s implicit response is that availability is not the same as access. An AI assistant that cannot be triggered by voice, cannot read what is on the screen, and cannot interact with the operating system’s core apps is not competing on equal terms, regardless of whether it is available for download.

The timing is not coincidental

Google delayed the full transition from Google Assistant to Gemini on Android from 2025 to 2026, with the final Assistant shutdown on mobile targeted for March 2026. The regulatory proceedings opened the same month. As Google completes the process of making Gemini the default AI experience on every Android phone, the Commission is simultaneously defining the terms on which rivals must be allowed to occupy that same position. The two timelines are on a collision course, with the binding decision due in July and Gemini’s entrenchment deepening with every software update.

The broader DMA enforcement picture adds pressure. The Commission already found Google in breach of DMA obligations regarding search self-preferencing in 2024 and opened separate non-compliance proceedings over the Play Store’s anti-steering rules. The competition concerns raised by Google’s AI partnerships, including the UK Competition and Markets Authority’s investigation into Google’s $2 billion investment in Anthropic, suggest that regulators across jurisdictions see Google’s position in AI as an extension of its existing market power rather than a fresh competitive start.

Meanwhile, the Court of Justice of the European Union is expected to rule on Google’s appeal of the original 2018 Android antitrust fine, reduced to €4.125 billion by a lower court in 2022. In June 2025, the court’s Advocate General recommended rejecting the appeal. If the CJEU upholds the fine, it will confirm as settled law the principle that Google illegally tied its services to Android, the same principle the DMA now codifies as a forward-looking obligation. The old case and the new proceedings are bookends of the same argument, separated by eight years and the arrival of AI.

Apple is watching

Google is not the only gatekeeper navigating DMA obligations around AI assistants. Apple was hit hard by EU rules and delayed its Apple Intelligence suite in Europe over DMA interoperability concerns. In response, Apple began allowing EU users to set a default voice assistant other than Siri under iOS 26.2, and Bloomberg reported in March that Apple plans to open Siri to rival AI services beyond its existing ChatGPT partnership in iOS 27. The pattern is consistent: both platform owners are being forced to treat AI assistants as a contestable layer rather than a proprietary feature.

The difference is that Apple is moving pre-emptively, however reluctantly, while Google is arguing that the requirements are unnecessary. That strategic divergence may matter. The DMA gives the Commission the power to impose fines of up to 10% of global annual turnover for non-compliance, rising to 20% for repeat offenders. For Alphabet, 10% of turnover would exceed $30 billion. The specification proceedings are technically neutral on compliance, the Commission frames them as “assisting” Google in meeting its obligations, but the enforcement machinery behind them is not.

The stakes beyond Europe

The debate over whether EU regulation helps or hinders AI competition is not settled. European AI startups raised $52 billion in 2024 against $209 billion in the United States, a gap that critics of the DMA argue regulation will widen. The counterargument, which the Commission is making in practice if not always in rhetoric, is that a market in which Google, Apple, and Microsoft can embed their AI assistants at the operating system level while rivals are confined to app stores is not a market that will produce European AI champions regardless of how much venture capital is available. Access to the platform is a precondition for competition, not a barrier to innovation.

The tension between regulation and competitiveness is real. The Commission’s Digital Omnibus package has proposed amendments to the AI Act and GDPR, acknowledging that Europe’s regulatory framework may need loosening to keep pace with American and Chinese AI development. But the DMA proceedings against Google represent a different bet: that the problem is not too much regulation but too little enforcement of the regulation that exists. If Google can make Gemini the default intelligence layer of two billion phones without giving rivals equivalent access, the AI market will be decided not by which model is best but by which company owns the operating system. The Commission has until July to decide whether that outcome is acceptable. The draft findings suggest it is not.