Malicious Notifications Could Trick Google Gemini Users

A prompt injection flaw in Google Gemini's voice assistant let attackers hide malicious commands in notifications, enabling social engineering and more.