EU Mandates Google Open Android AI Services and Search Data to Competitors
Key Takeaways
- ▸Google must grant third-party AI assistants system-level access to Android features including hardware buttons, screen capture, cameras, microphones, and always-on concurrent wake-word detection from multiple AIs simultaneously
- ▸Third-party AIs gain equal access to Google's on-device AI models and all data from Google's own apps (Gmail, Maps, Calendar, Docs) that Gemini currently accesses
- ▸Google is ordered to share anonymized search data with competitors under FRAND pricing, including search terms, clicks, device types, and languages; Google bears responsibility for filtering personally-identifying information
Summary
The European Commission has issued two binding specification measures to Google under the Digital Markets Act, aimed at reducing its dominance in AI and search services. The first requires Google to grant third-party AI assistants equal access to Android device features currently exclusive to Google's Gemini, including hardware button control, screen capture, microphone and camera access, background execution, and always-on wake-word detection capabilities. The second mandates that Google share anonymized search query data with competitors at fair, reasonable, and non-discriminatory (FRAND) prices.
The Android AI interoperability requirement is remarkably expansive in scope. Google must provide third-party AI services with access to its on-device local models and all information from Google's own applications (Gmail, Calendar, Docs, Maps, etc.) that Gemini currently accesses. Critically, the guidance requires Google to permit concurrent always-on audio detection from multiple AI assistants simultaneously—meaning users could have ChatGPT, Claude, Grok, and other third-party AIs all listening for their wake words at the same time alongside Gemini. The guidance also prohibits third-party app developers from exclusive AI partnerships; if Slack enables AI access to its data, it must work equally with all system-level AI providers.
The search data sharing measure addresses Google's data advantage by forcing it to share search behavior insights—including query terms, user clicks, device types, and languages—with all competitors. While Google can charge for access, pricing must follow Commission-defined FRAND methodology. The Commission places responsibility on Google to filter out personally-identifying information like passwords and usernames before sharing datasets.
These measures represent an unprecedented regulatory intervention into how platforms manage AI systems and competitive access, raising significant questions about privacy, security risks from concurrent device access, and the practical feasibility of managing system-level interoperability at this scale.
- Third-party app developers cannot create exclusive AI partnerships—if they enable AI access to their data, it must be available to all system-level AI providers equally
