Alexa, Siri, Google Assistant vulnerable to malicious commands, study reveals

Amazon researchers uncover alarming vulnerabilities in speech AI models, revealing 90% susceptibility to adversarial attacks and raising concerns over potential misuse.