Google is upgrading its Gmail service with enhanced AI features, impacting its 2 billion users worldwide. The update will allow Google’s cloud-based AI to access users’ content, including sensitive data, raising privacy concerns.
Previously, Gmail’s AI features conflicted with its limited end-to-end encryption, posing a privacy challenge. Similar issues emerged this week for Android users when an unclear email suggested Google’s Gemini AI would access sensitive apps like Messages and WhatsApp even if users disabled Gemini Activity.
Google clarified the confusion, stating that Gemini can assist with phone functions such as messaging, calls, and timers regardless of Gemini Apps Activity settings. However, with Gemini Apps Activity turned off, chats are not reviewed or used to improve AI models. Previously, users had to keep Gemini Activity on for these features, which meant interactions were stored—a notable privacy concern that Google now addresses.
Despite these clarifications, Android users should note that Gemini interactions are still stored temporarily for 72 hours, even if activity tracking is disabled.
Starting July 7, 2025, Gemini AI will access apps like Phone and Messages irrespective of Gemini Apps Activity settings. While data won’t be used for AI training when the setting is off, this deep integration raises privacy concerns about temporary access to personal data such as call logs and private messages.
Similar AI integrations are appearing on other platforms, with WhatsApp planning AI summaries of message threads, highlighting the ongoing balance between user convenience and data privacy.
Industry reports emphasize that most users lack full understanding of complex privacy policies associated with AI features, especially younger audiences. The rollout of Gemini AI in education settings further intensifies privacy discussions, as schools deploy AI tools for lesson planning and feedback.
Incogni’s 2025 “Gen AI and LLM Data Privacy Ranking” warns of the rapid data exposure risks tied to AI tools like Gemini, Microsoft’s Copilot, and others embedded into widely used operating systems. Meta ranks lowest in privacy practices, with Google placed near the bottom but ahead of Microsoft. ChatGPT scores higher, feeding narratives that users may prefer independent AI platforms over big tech’s offerings.
While maintaining awareness of evolving privacy risks is challenging for most users, Google asserts strong privacy protocols for Gemini, particularly in education. According to TechRadar, Gemini for Education adheres to the same protections as Workspace for Education, ensuring student data isn’t used for AI training or human review.