Hackers Find That Inaudible Sounds Hidden in Podcasts or Random Videos Can Hijack Your AI Voice Chatbot
TL;DR
Researchers have shown that inaudible ultrasonic sounds hidden in podcasts or random videos can secretly issue commands to AI voice chatbots and smart speakers. The attack exploits weaknesses in voice-command pipelines, letting third parties trigger actions on a device without the user's consent. The risk grows for anyone running always-on voice assistants in the background. Practical mitigation includes tighter microphone permissions and prompt firmware updates from vendors.
Nauti's Take
Real upside in this research: every disclosed attack vector forces vendors to harden voice-AI pipelines that have stayed sloppy for years. The catch is brutal — inaudible commands buried in random podcasts or background video make smart speakers practically impossible to defend at the user layer.
Anyone running always-on mics should tighten permissions hard; the real fix has to come from vendors making ultrasonic filtering a default expectation, not a checkbox feature.
Sources