The European Commission has ordered Google to open 11 Android operating-system features to competing AI developers by July 2027 and to share anonymized Google Search data with qualifying rivals by January 2027, turning the Digital Markets Act into a direct challenge to Gemini’s home-field advantage. For Europeans using Android—and for the developers trying to build alternatives to Google Search and Google’s assistant stack—the decision matters because it targets the two advantages that are hardest to recreate from outside Google: deep access to the phone and access to search data collected at enormous scale.
This is not simply another EU complaint about app-store rules, browser defaults, or preinstalled software. The Commission is asserting that the AI assistant race cannot be treated as a clean-sheet market when one contestant already controls the operating system, the default services, the voice trigger, and the search engine that supplies much of the web’s discovery layer. Google may still make Gemini the best assistant, but it will no longer be allowed to reserve the underlying Android pathways solely for its own service in the EU.
The ruling is therefore a test of whether the DMA can stop platform power from becoming AI power before the market settles. Google sees a privacy and security risk. Brussels sees a competition problem with privacy controls built into the remedy. Both positions contain a real warning for users and IT administrators—but the Commission’s central point is harder to dismiss: an assistant that cannot wake on command, understand device context, or perform system-level actions is not meaningfully competing with one that can.
The Commission detailed Google’s obligations on Thursday, six months after launching specification proceedings under the DMA. The timing is important. These are not retrospective penalties designed only to punish conduct that has already happened; they are implementation rules intended to define what effective competition must look like while AI assistants are still becoming central to how people use phones and search for information.
As reported by Tekedia, the Android portion requires Google to make 11 operating-system features available to competing AI developers. The practical effect is that a third-party assistant should be able to participate in the kinds of interactions users currently associate with Google Assistant and Gemini: responding to a voice invocation, searching for local information, booking transportation, and carrying out other system-level tasks.
PCMag framed the disparity in unusually direct terms. The Commission said competitors’ assistants currently have only restricted access to key Android functions, while Google’s own AI services have full access. A rival tool may be competent at answering questions in a chat window, but if it cannot be launched as naturally as “Hey Google,” cannot work across applications, and cannot act on user requests within the operating system, then its theoretical intelligence does not translate into an equivalent product.
Computing UK’s example—booking a taxi—captures the distinction. Generating a useful answer is one thing. Completing an action across the device, applications, permissions, and account context is another. That is the moment where an AI chatbot becomes an assistant, and it is also where a platform owner can quietly turn technical integration into an enduring competitive moat.
The Commission’s own estimate that 60% of EU users have an Android device makes the issue more than a niche platform dispute. A large population may be able to download rival assistants, but downloading is not the same as being offered an equal chance to become the phone’s everyday interface.
Google retains more than 91% of the global search market, according to PCMag. That dominance is not just a matter of brand recognition or a preferred browser setting. Search improves through the continuous accumulation of ranking, query, click, and view signals: what people seek, which results they select, how they revise a search, and what satisfies a query. Those signals help make a search engine useful long before a consumer decides which interface they like best.
The Commission’s demand, as summarized by Computing UK, is that Google share the same data it collects to optimize its own search services, subject to anonymisation. The order also treats AI chatbots as search services for the purpose of data sharing. That detail matters enormously. It means the framework is not limited to old-style web search engines with a results page; it recognizes that AI products which retrieve and synthesize information are increasingly part of the same discovery market.
This could give companies such as OpenAI, identified by Tekedia as a potential beneficiary, a route to improve AI-powered search products without having to begin with Google’s two-decade data advantage. It does not hand rivals Google Search, its index, or its consumer relationships. Nor does it guarantee that smaller firms will turn the data into better products. But it changes a basic economic equation: competition will no longer depend entirely on whether a newcomer can independently collect enough user behavior at scale before its funding, patience, or relevance runs out.
That is why the search remedy is more than a technical data-access obligation. It is a regulatory judgment that Google’s feedback loop is itself a gatekeeping asset. In the conventional search era, the concern was Google favoring its own vertical products. In the AI era, the concern is that Google can convert its search scale into a self-reinforcing lead for Gemini and any AI-inflected version of Search.
Computing UK reported Walker’s related warning that the approach threatens Android device security and could expose private searches to unfamiliar companies. This is the strongest version of Google’s case: the more deeply an assistant can access device functions, context, and user intent, the more damaging a bad implementation or a malicious developer could be. A competing assistant that can invoke itself by voice and execute actions is not merely another app asking for notification permission.
But the Commission’s measures do not describe a universal entitlement for every AI app to acquire privileged system access. Tekedia reports that developers must meet specific privacy and cybersecurity requirements before gaining the Android capabilities. On search data, Google may assess whether an applicant poses cybersecurity or data-protection risks before sharing information. The data-sharing system also requires anonymization and uses a commercial-access framework rather than an instruction to publish Google’s internal data freely.
That does not eliminate risk. Anonymization is difficult, security review is only as good as its execution, and the most powerful AI experiences increasingly depend on data flows that users may struggle to understand. Still, Google’s argument cannot be allowed to become a blanket defense of exclusive access. If only the platform owner can ever be trusted with the privileged hooks that make an assistant useful, then “security” becomes a permanent synonym for “incumbent control.”
The more revealing question is whether the safeguards will be administered fairly. Google retains a risk-assessment role, which is rational given the potential damage of weak access controls. Yet a gatekeeper evaluating the eligibility of companies that want to compete with its flagship AI service creates an obvious tension. The Commission will need to watch not only whether Google creates APIs and data-sharing channels, but whether qualification, review, pricing, documentation, and timing make those channels genuinely usable.
July 17, 2026: Tekedia, PCMag, and Computing UK reported the decision’s significance for Gemini rivals, search competitors, privacy, and Android access.
January 2027: Google must implement the search-data sharing requirements.
July 2027: Android AI integration changes are scheduled to become available with the relevant Android release.
The Android case has long revolved around how Google’s mobile platform could reinforce Google services such as Chrome and Google Search. The present DMA action extends that familiar theory into a new layer of computing. If Android once helped determine which browser and search engine a user encountered, the next fight is over which assistant hears the request, sees the relevant context, makes the recommendation, and completes the action.
That is why the ruling should not be misread as an EU endorsement of any particular Google rival. Brussels is not declaring that OpenAI or another company deserves to replace Gemini. It is trying to ensure they can compete without being structurally hobbled by Android’s access rules or by the lack of data that Google alone has accumulated.
Henna Virkkunen, the EU Executive Vice President for Tech Sovereignty, Security and Democracy, put the political goal plainly: the Commission hopes to see emerging alternatives to Google Search and Google AI services such as Gemini, so EU users gain greater choice. This is less a verdict on the quality of Gemini than a verdict on the conditions under which alternatives can prove their quality.
The future question will not simply be whether an AI app is permitted on a phone. It will be which assistant can accept a voice trigger, what work-related context it can reach, which tasks it can execute, where its data is processed, and how its permissions interact with existing Android enterprise controls. In practice, the DMA’s promise of choice means administrators may face more technically credible assistant options—and more pressure to make principled decisions about them.
Google’s security warning should therefore be treated as a call for disciplined deployment, not as a reason to preserve the status quo indefinitely. The Commission is requiring access only for developers that meet privacy and cybersecurity conditions, but employers will still need their own vendor review, data-handling rules, identity controls, and incident-response assumptions. “Eligible under the DMA” will not automatically mean “approved for corporate use.”
The deadlines also remove a convenient excuse for everyone involved. Google cannot say the Commission demanded instant, reckless exposure of its systems. Competitors cannot say Android access was impossible forever. By 2027, the question will become concrete: did Google provide effective, fair access, and did alternative AI services use it to produce experiences consumers actually prefer?
The requirements reportedly extend to equal access for eligible rivals to relevant on-device system models, including Gemini Nano where those models are part of Google’s DMA-designated Android operating system. The specifications also cover invocation mechanisms, consent-based access to device context, and structured actions across Android and participating apps.
One important exception has a later deadline. Concurrent hotword detection, intended to prevent Google from reserving voice wake-word access for its own assistant, can be deferred to Android 19, due no later than August 1, 2028.
For IT teams, the clarification reinforces that this will be a staged platform change rather than an immediate policy switch. Android Enterprise, work-profile controls, permission management, and vendor review processes will need to account for assistants that may gain deeper access to screen context, microphone input, background execution, and app actions.
The report also describes more specific privacy controls: direct identifiers would be removed, rare queries suppressed, and records grouped with at least 1,000 users sharing language, region, and device-class characteristics. Independent annual audits are expected to test those protections.
Notebookcheck further reports that Google is expected to appeal, but that an appeal would not suspend its obligation to comply.
This is not simply another EU complaint about app-store rules, browser defaults, or preinstalled software. The Commission is asserting that the AI assistant race cannot be treated as a clean-sheet market when one contestant already controls the operating system, the default services, the voice trigger, and the search engine that supplies much of the web’s discovery layer. Google may still make Gemini the best assistant, but it will no longer be allowed to reserve the underlying Android pathways solely for its own service in the EU.
The ruling is therefore a test of whether the DMA can stop platform power from becoming AI power before the market settles. Google sees a privacy and security risk. Brussels sees a competition problem with privacy controls built into the remedy. Both positions contain a real warning for users and IT administrators—but the Commission’s central point is harder to dismiss: an assistant that cannot wake on command, understand device context, or perform system-level actions is not meaningfully competing with one that can.
Brussels Has Moved the Fight From Defaults to Capabilities
The Commission detailed Google’s obligations on Thursday, six months after launching specification proceedings under the DMA. The timing is important. These are not retrospective penalties designed only to punish conduct that has already happened; they are implementation rules intended to define what effective competition must look like while AI assistants are still becoming central to how people use phones and search for information.As reported by Tekedia, the Android portion requires Google to make 11 operating-system features available to competing AI developers. The practical effect is that a third-party assistant should be able to participate in the kinds of interactions users currently associate with Google Assistant and Gemini: responding to a voice invocation, searching for local information, booking transportation, and carrying out other system-level tasks.
PCMag framed the disparity in unusually direct terms. The Commission said competitors’ assistants currently have only restricted access to key Android functions, while Google’s own AI services have full access. A rival tool may be competent at answering questions in a chat window, but if it cannot be launched as naturally as “Hey Google,” cannot work across applications, and cannot act on user requests within the operating system, then its theoretical intelligence does not translate into an equivalent product.
Computing UK’s example—booking a taxi—captures the distinction. Generating a useful answer is one thing. Completing an action across the device, applications, permissions, and account context is another. That is the moment where an AI chatbot becomes an assistant, and it is also where a platform owner can quietly turn technical integration into an enduring competitive moat.
The Commission’s own estimate that 60% of EU users have an Android device makes the issue more than a niche platform dispute. A large population may be able to download rival assistants, but downloading is not the same as being offered an equal chance to become the phone’s everyday interface.
| Area | Google’s current advantage, according to the Commission | Required change | Implementation deadline | Built-in condition |
|---|---|---|---|---|
| Android AI assistants | Gemini and Google services have fuller access to key Android functions | Open 11 Android features to eligible competing AI developers | July 2027 | Developers must meet privacy and cybersecurity requirements |
| Voice and system actions | Third-party assistants have restricted ability to invoke and complete device-level tasks | Enable rival assistants to support deeper integration, including voice activation and system tasks | July 2027 | Access is not unrestricted |
| Google Search data | Google uses data collected to optimize its own search services | Share relevant anonymized data with eligible competitors, including AI chatbots offering search | January 2027 | Google can assess cybersecurity and data-protection risk |
| Commercial access | Google controls the data pipeline behind search optimization | Offer access transparently and for a reasonable fee under a set formula | January 2027 | Data must be anonymized |
Search Data Is the More Radical Part of the Order
The Android requirements will attract the biggest headlines because they change what users can see and do on a phone. The Search decision may prove more consequential over time because it strikes at the information advantage behind the services.Google retains more than 91% of the global search market, according to PCMag. That dominance is not just a matter of brand recognition or a preferred browser setting. Search improves through the continuous accumulation of ranking, query, click, and view signals: what people seek, which results they select, how they revise a search, and what satisfies a query. Those signals help make a search engine useful long before a consumer decides which interface they like best.
The Commission’s demand, as summarized by Computing UK, is that Google share the same data it collects to optimize its own search services, subject to anonymisation. The order also treats AI chatbots as search services for the purpose of data sharing. That detail matters enormously. It means the framework is not limited to old-style web search engines with a results page; it recognizes that AI products which retrieve and synthesize information are increasingly part of the same discovery market.
This could give companies such as OpenAI, identified by Tekedia as a potential beneficiary, a route to improve AI-powered search products without having to begin with Google’s two-decade data advantage. It does not hand rivals Google Search, its index, or its consumer relationships. Nor does it guarantee that smaller firms will turn the data into better products. But it changes a basic economic equation: competition will no longer depend entirely on whether a newcomer can independently collect enough user behavior at scale before its funding, patience, or relevance runs out.
That is why the search remedy is more than a technical data-access obligation. It is a regulatory judgment that Google’s feedback loop is itself a gatekeeping asset. In the conventional search era, the concern was Google favoring its own vertical products. In the AI era, the concern is that Google can convert its search scale into a self-reinforcing lead for Gemini and any AI-inflected version of Search.
The Commission Is Not Ordering an Unlocked Android Free-for-All
Google’s objection is neither surprising nor trivial. Kent Walker, Google’s president of global affairs and chief legal officer, said the decisions risk “undermining vital privacy and security guardrails for millions of Europeans.” He also said Google had repeatedly offered solutions intended to satisfy the DMA while protecting users, but that the rulings discounted evidence of user harm.Computing UK reported Walker’s related warning that the approach threatens Android device security and could expose private searches to unfamiliar companies. This is the strongest version of Google’s case: the more deeply an assistant can access device functions, context, and user intent, the more damaging a bad implementation or a malicious developer could be. A competing assistant that can invoke itself by voice and execute actions is not merely another app asking for notification permission.
But the Commission’s measures do not describe a universal entitlement for every AI app to acquire privileged system access. Tekedia reports that developers must meet specific privacy and cybersecurity requirements before gaining the Android capabilities. On search data, Google may assess whether an applicant poses cybersecurity or data-protection risks before sharing information. The data-sharing system also requires anonymization and uses a commercial-access framework rather than an instruction to publish Google’s internal data freely.
That does not eliminate risk. Anonymization is difficult, security review is only as good as its execution, and the most powerful AI experiences increasingly depend on data flows that users may struggle to understand. Still, Google’s argument cannot be allowed to become a blanket defense of exclusive access. If only the platform owner can ever be trusted with the privileged hooks that make an assistant useful, then “security” becomes a permanent synonym for “incumbent control.”
The more revealing question is whether the safeguards will be administered fairly. Google retains a risk-assessment role, which is rational given the potential damage of weak access controls. Yet a gatekeeper evaluating the eligibility of companies that want to compete with its flagship AI service creates an obvious tension. The Commission will need to watch not only whether Google creates APIs and data-sharing channels, but whether qualification, review, pricing, documentation, and timing make those channels genuinely usable.
Timeline
July 16, 2026: The European Commission detailed binding DMA specifications for Google’s Android AI interoperability and Google Search data-sharing obligations.July 17, 2026: Tekedia, PCMag, and Computing UK reported the decision’s significance for Gemini rivals, search competitors, privacy, and Android access.
January 2027: Google must implement the search-data sharing requirements.
July 2027: Android AI integration changes are scheduled to become available with the relevant Android release.
Google’s Old Antitrust Record Explains Why the EU Is Acting Early
The Commission is not approaching Google’s AI strategy in a vacuum. Computing UK notes that Google has already faced a €4.1 billion penalty over anticompetitive Android practices, while further fines totaling €5.35 billion concerned unfairly favoring Google’s own products in other markets, including shopping comparison services. The contrast with Apple and Meta’s combined €700 million in fines illustrates the scale of the competition history surrounding Google’s ecosystem.The Android case has long revolved around how Google’s mobile platform could reinforce Google services such as Chrome and Google Search. The present DMA action extends that familiar theory into a new layer of computing. If Android once helped determine which browser and search engine a user encountered, the next fight is over which assistant hears the request, sees the relevant context, makes the recommendation, and completes the action.
That is why the ruling should not be misread as an EU endorsement of any particular Google rival. Brussels is not declaring that OpenAI or another company deserves to replace Gemini. It is trying to ensure they can compete without being structurally hobbled by Android’s access rules or by the lack of data that Google alone has accumulated.
Henna Virkkunen, the EU Executive Vice President for Tech Sovereignty, Security and Democracy, put the political goal plainly: the Commission hopes to see emerging alternatives to Google Search and Google AI services such as Gemini, so EU users gain greater choice. This is less a verdict on the quality of Gemini than a verdict on the conditions under which alternatives can prove their quality.
Enterprise Android Management Will Become an AI Governance Problem
For Windows and enterprise IT readers, this is an Android story with a cross-platform consequence. Organizations increasingly manage a mixed environment: Windows endpoints, Android phones, cloud identity, mobile device management, corporate data in browser sessions, and employees asking AI tools to search, summarize, draft, and act. A more capable third-party Android assistant changes the boundary of what must be assessed before an employee installs or adopts it.The future question will not simply be whether an AI app is permitted on a phone. It will be which assistant can accept a voice trigger, what work-related context it can reach, which tasks it can execute, where its data is processed, and how its permissions interact with existing Android enterprise controls. In practice, the DMA’s promise of choice means administrators may face more technically credible assistant options—and more pressure to make principled decisions about them.
Google’s security warning should therefore be treated as a call for disciplined deployment, not as a reason to preserve the status quo indefinitely. The Commission is requiring access only for developers that meet privacy and cybersecurity conditions, but employers will still need their own vendor review, data-handling rules, identity controls, and incident-response assumptions. “Eligible under the DMA” will not automatically mean “approved for corporate use.”
Action checklist for admins
- Inventory which Android AI assistants are currently allowed, blocked, or unmanaged across corporate and BYOD fleets.
- Review whether existing mobile policies distinguish a simple chatbot from an assistant able to invoke by voice or carry out system-level tasks.
- Update AI vendor assessments to cover data retention, cross-app actions, account binding, and handling of potentially sensitive search or workplace context.
- Monitor Android management and security documentation ahead of the July 2027 integration deadline.
- Prepare user guidance that explains why voice-enabled and action-capable assistants require a higher level of scrutiny than ordinary productivity apps.
The Dates Give Google Time—And Rivals a Test They Cannot Dodge
The staggered schedule is telling. Search-data sharing arrives first, in January 2027; deeper Android changes follow in July 2027. Google gets time to build controls, establish eligibility processes, and operationalize anonymization. Rivals get time to decide whether they can turn the new access into compelling products rather than merely demand parity in principle.The deadlines also remove a convenient excuse for everyone involved. Google cannot say the Commission demanded instant, reckless exposure of its systems. Competitors cannot say Android access was impossible forever. By 2027, the question will become concrete: did Google provide effective, fair access, and did alternative AI services use it to produce experiences consumers actually prefer?
What will matter when the 2027 deadlines arrive
The order is ambitious, but its value will be determined by implementation rather than rhetoric.- Eleven Android features must become accessible to eligible rival AI developers.
- January 2027 is the deadline for anonymized Google Search data sharing.
- July 2027 is the target for Android AI interoperability changes.
- AI chatbots count as search services when they offer search functionality under the data-sharing rules.
- Privacy and cybersecurity controls remain central, including developer eligibility and Google’s ability to assess access risks.
- Gemini remains Google’s product to beat, but its privileged platform position is the part Brussels is trying to narrow.
Update: Android 18 deadline set for August 1, 2027 (July 18, 2026)
New reporting from Simply Wall St adds a more specific implementation timetable to the Commission’s Android interoperability order. Most of Google’s required AI-assistant changes must be delivered in Android 18 by August 1, 2027—not simply at an unspecified point in July 2027.The requirements reportedly extend to equal access for eligible rivals to relevant on-device system models, including Gemini Nano where those models are part of Google’s DMA-designated Android operating system. The specifications also cover invocation mechanisms, consent-based access to device context, and structured actions across Android and participating apps.
One important exception has a later deadline. Concurrent hotword detection, intended to prevent Google from reserving voice wake-word access for its own assistant, can be deferred to Android 19, due no later than August 1, 2028.
For IT teams, the clarification reinforces that this will be a staged platform change rather than an immediate policy switch. Android Enterprise, work-profile controls, permission management, and vendor review processes will need to account for assistants that may gain deeper access to screen context, microphone input, background execution, and app actions.
Alphabet (GOOGL) Faces EU Android Order As Gemini 3.5 Pro Slips - Simply Wall St News
EU regulators required Google to open parts of the Android operating system and anonymized search data to rival AI providers under the Digital Markets Act.
The new rules target Google’s position in mobile and search based services across the EU market.
At the same time, Alphabet reported...
simplywall.st
Update: Google Search data package reportedly due in November 2026 (July 18, 2026)
Notebookcheck reports that Google must have the anonymized Search dataset ready by November 2026, ahead of the January 2027 deadline for the commercial-access pricing model. That adds an earlier operational milestone to the Commission’s search-data remedy.The report also describes more specific privacy controls: direct identifiers would be removed, rare queries suppressed, and records grouped with at least 1,000 users sharing language, region, and device-class characteristics. Independent annual audits are expected to test those protections.
Notebookcheck further reports that Google is expected to appeal, but that an appeal would not suspend its obligation to comply.
References
- Primary source: Tekedia
Published: 2026-07-17T23:54:11+00:00
EU Orders Google To Open Android And Search To AI Rivals, Reshaping Competition With Gemini - Tekedia
Alphabet’s Google will be required to give artificial intelligence rivals, including OpenAI, and competing search providers greater access to key Android and Search services under new European Union rules designed to curb the market power of Big Tech and foster greater competition in the...www.tekedia.com - Independent coverage: PCMag
Published: 2026-07-17T14:02:20+00:00
EU Will Force Google to Open Up Android to Rival AI Chatbots | PCMag
It also has to reveal its internal search data to competitors. Google, however, argues that both moves would harm the privacy and security of its users.www.pcmag.com - Independent coverage: Computing UK
Published: Fri, 17 Jul 2026 10:10:55 GMT
EU forces Google to open AI and Search products
The European Union is legally compelling Google to open its ecosystem to competitors with changes to how it handles AI and search.www.computing.co.uk - Related coverage: digital-markets-act.ec.europa.eu
Alphabet specification proceedings - Sharing of Google Search data
Alphabet Google Search specification proceedingsdigital-markets-act.ec.europa.eu - Related coverage: digital-strategy.ec.europa.eu
Commission opens proceedings to assist Google in complying with interoperability and online search data sharing obligations under the Digital Markets Act
The European Commission has started two sets of specification proceedings to assist Google in complying with its obligations under the Digital Markets Act (‘DMA').digital-strategy.ec.europa.eu
- Related coverage: computerworld.com
Google must open Android to rival AI agents, EU orders – Computerworld
The move has implications for CISOs as well, says an industry expert.
www.computerworld.com
Last edited:
