Google accused of using Gemini AI to track private communications of users

Gemini

Google is facing serious allegations that its Gemini AI assistant secretly accessed users’ private communications across Gmail, Chat, and Meet. A new lawsuit filed in a California federal court claims that Google unlawfully enabled Gemini to collect personal data without the knowledge or consent of users, as per a report by Bloomberg.

According to the complaint, filed in San Jose, Google “quietly” turned on Gemini across its popular communication apps in October. Previously, users were given the option to activate the AI assistant, but the suit alleges that the tech giant made the switch automatic — effectively allowing Gemini to analyze private messages, emails, and attachments behind the scenes.

The lawsuit, which seeks class-action status, argues that this move violates the California Invasion of Privacy Act, a state law that bans recording or eavesdropping on private conversations without permission from all parties involved. The filing states that unless users manually turn Gemini off in Google’s privacy settings — which requires digging through several menus — the AI tool continues to “access and exploit” users’ data in the background.

In simpler terms, the complaint claims that Gemini is effectively reading every email and message users send or receive through Google’s services. That includes text, attachments, and even files shared in conversations.

Google has not yet issued a public comment on the lawsuit. However, privacy experts say the case could reignite concerns over how AI models are being trained and whether they are crossing ethical or legal lines in how they collect information.

Gemini is Google’s advanced AI system, built to help users write, summarize, and generate content directly inside apps like Gmail and Docs. While the company says its AI tools are designed to assist and not spy, the lawsuit paints a different picture — one where convenience comes at the cost of user privacy.

The case, Thele v. Google LLC (25-cv-09704), will be heard in the U.S. District Court for the Northern District of California. If proven, it could become one of the most significant privacy battles involving AI to date.