EU Mandates Google Open Android Access for Rival AI Assistants

The European Commission has issued a sweeping mandate for Google to open up core functionalities of its Android operating system to competing AI assistants. Effective with the next major Android release, anticipated as Android 18, and no later than August 1, 2027, rival AI services will gain access to elements previously exclusive to Google’s own offerings, including the camera, microphone, and on-screen content. This decision, adopted under the Digital Markets Act (DMA), also compels Google to share anonymized search query data with competitors for a fee.

This regulatory action represents a significant step towards leveling the playing field in the increasingly competitive landscape of AI-powered mobile assistants. For years, Google’s proprietary access to Android’s deep system-level features has given its own AI, like Gemini, a distinct advantage. The new rules aim to dismantle these barriers, allowing third-party AI assistants to offer comparable functionality, such as background operation, wake-word activation even when the screen is off, and the ability to mimic user interactions like taps and typing.

Enhanced Access for Third-Party AI

The European Commission’s decision specifies eleven operating system features, with Google permitted to restrict access to five designated as “restricted features.” These include centralized on-device data access via AppSearch, context-aware intelligence systems like Magic Cue, structured on-device integration for actions and functions, screen automation through Computer Control, and system integration for settings, media, screenshots, notifications, and power management. Certified assistants will require permission to access these functionalities, which can include drafting emails, managing calendars, controlling media playback, and reading/writing messages.

However, six other features will be open to all third-party applications without certification requirements. These include ambient data collection (microphone, audio, camera, screen, location, sensors), always-on hotword detection, long-press invocation, on-device AI models, third-party model implementation, and background execution. These open features will operate under similar user consent prompts currently used by Google's first-party services, ensuring that user permission remains paramount. Ambient data collection, for instance, will now be accessible continuously and in the background, mirroring the lighter consent processes Google’s own services utilize.

The ruling also addresses specific functionalities like simultaneous listening by multiple assistants, slated for Android 19 by August 2028. Google can still implement process isolation and encryption, but it cannot dictate which apps are allowed to request access to these newly opened features. If Google wishes to gate a feature, it must file a formal request with the Commission outlining good cause. For the restricted features, Google is required to establish a “Qualified AI Assistant Programme.” This program will allow third-party companies to have their AI assistants certified by independent Trusted Certification Authorities (TCAs) free of charge. Google cannot impose additional conditions or arbitrarily revoke certifications, though it can set reasonable terms for TCAs and test assistants for security and intent confirmation before sensitive actions.

The second part of the Commission’s ruling targets Google's dominance in search. Beginning in 2027, Google must provide anonymized search query, click, and ranking data to rival search engines and AI chatbots that perform search functions. The anonymization process involves multiple passes to strip identifiers, suppress sensitive queries, and group users into large cohorts based on shared characteristics like location and device type. Recipients must meet strict criteria, including having at least 50,000 average monthly EU users and not being controlled by entities deemed high-risk by the EU. This data sharing aims to fuel competition in the search and AI discovery space.

Google, through its President of Global Affairs Kent Walker, expressed concerns, stating the Android decision “threatens device security by granting external apps sensitive and powerful device permissions” and that it removes safeguards currently employed by phone manufacturers. Regarding the search data, Google argues the anonymization is insufficient and that user consent has not been adequately obtained, potentially exposing trade secrets and impacting national security. The company cited a recent paper from ENISA, the EU’s cybersecurity agency, emphasizing the importance of security fundamentals in the AI era, though the paper did not specifically address Android interoperability.

Despite Google's objections, the European Commission’s directives under the Digital Markets Act are binding. The mandates are designed to curb the gatekeeper power of large tech platforms and foster a more competitive digital environment within the European Union. The implications extend beyond the EU, potentially influencing regulatory approaches in other regions grappling with the dominance of Big Tech in the AI and mobile operating system markets.