Security researchers have uncovered a critical vulnerability in Google’s Gemini AI, allowing attackers to gain control of smart home devices through manipulated Google Calendar invites. Dubbed ‘promptware,’ the attack exploits Gemini’s integration with other Google services to inject malicious commands via calendar events. This allows remote control of connected devices. Security publications Wired and Ars Technica have reported on the findings. Further details can be found on Reddit.