Google sued over claims Gemini AI tracked private user data

Google is facing a lawsuit in the US that claims its Gemini AI assistant unlawfully tracked private communications of users on Gmail, Chat, and Meet.

The suit, filed in federal court in San Jose, alleges Google enabled Gemini by default for these services in October 2025, allowing it to collect user data without explicit consent.

The complaint says users must navigate privacy settings to turn Gemini off, and claims the tool accessed emails, attachments, and other communications.

Plaintiffs argue this violates the California Invasion of Privacy Act, which bans recording confidential communications without all parties’ consent.

The case is being brought as a proposed class action.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

Gemini defaults vary; all-party consent creates the risk

Support pages say Ask Gemini in Meet is on by default unless an organization’s IT admin disables it 1. In Google Workspace services (Google’s paid productivity suite for businesses), Gemini features are on for supported editions, with admins able to turn them off 2.
California’s Invasion of Privacy Act (CIPA) requires explicit all-party consent before recording or intercepting communications 3. Ask Gemini in Meet reads live captions to create summaries and deletes them after the meeting 1. Workspace smart features can use data across Gmail, Chat, and Meet when enabled 4. The risk turns on whether any third party gets conversation content without prior consent, since a party exception may apply if processing stays inside Google 3.

Enterprise software vendors can address all-party consent gaps

The lawsuit exposes a gap for organizations that use AI across Meet, Zoom, and Microsoft Teams 3. Many lack tools to enforce and record all-party consent before AI processing begins. CIPA allows up to $5,000 per violation in class actions 3.
Consent management middleware (a software layer between systems) can intercept AI requests for transcription, summarization, or sentiment analysis. It can require opt in from all participants before data flows to models, with tamper evident logs for compliance. Business-to-business software-as-a-service (B2B SaaS) can focus on regulated buyers in healthcare, finance, or legal, since few tools handle real time consent for synchronous communications (live calls and meetings) with AI observers.