David Arroyo, a cybersecurity researcher, recently discovered that some interactions with smart speaker Alexa were recorded without the activation word, “Alexa”. This surprising revelation was shared in a viral video on TikTok and Instagram. Arroyo explains that false positives can sometimes occur in voice interpretation systems like those used by Alexa. These systems are designed to identify variations in pronunciation, accents, and room acoustics to improve accuracy.
Josep Albors from ESET Spain underscores the potential privacy risks associated with false positives in voice recognition systems. A study by researchers from Ruhr University Bochum and the Max Planck Institute for Security and Privacy highlighted the issue of accidental activations in smart speakers. They found over 1,000 involuntary activations in 11 devices from different manufacturers.
Amazon’s policy states that no audio is stored or sent to the cloud unless the device detects the wake word, ensuring user privacy. However, there is no evidence to suggest this is currently happening beyond keyword detection. While smart speakers like Alexa constantly monitor for activation words like “Alexa”, “ok, Google”, or “hey, Siri”, Amazon assures that their technology only relies on sound waves to detect these terms. Users can choose to opt-out of storing recordings in their privacy settings.
Arroyo and Albors agree that if sound data were processed beyond keyword detection, serious privacy concerns would arise. However, they emphasize the importance of ensuring user trust and confidence in smart devices to prevent potential economic damage to companies
