Google Gemini Voice Assistant Vulnerability Allowed Hijacking via Malicious Notifications

A security flaw in Google Gemini’s voice assistant on Android enabled malicious actors to hijack the assistant through a single poisoned notification from platforms like WhatsApp, Slack, SMS, Signal, Instagram, or Messenger. The attack did not require installing a malicious app, as the assistant processed harmful content directly from notifications. Exploits could trigger actions such as opening connected windows, fabricating messages from a victim’s boss, initiating Zoom calls, or corrupting the assistant’s long-term memory. The vulnerability affected Android devices running Google Gemini, though specific versions and disclosure dates were not provided. No CVE ID was mentioned in the reported details.