Always-on Mic-Enabled Devices Have Unique Privacy Implications, Paper Says

Always-on mic-enabled devices have unique privacy implications as compared with those that are manually or speech-activated, the Future of Privacy Forum and Ernst & Young said in a new paper released Thursday. The paper said it is inaccurate to label all such devices with speech recognition or mic-enabled features as being always on. Manually activated devices just require a user to press a button or flip a switch to record and transmit audio to a voice-to-text translation service, the paper said. Speech-activated devices like iPhone 6S or Microsoft Cortana stay in an "inert state of passive processing" and require a "wake phrase" to turn on. These devices are "not really 'listening' to its environment" but use the mic as another sensor. But always-on devices -- such as security cameras, baby monitors, the wristband Kapture or wearable camera OrCam -- "evoke different privacy concerns" from the other two categories, the paper said. Such devices "call for notice and consent frameworks in sync with the more extensive data collection that they enable," it added. The paper also discussed how some laws consider a "voice print" as a biometric or personal record with restrictions on usage. Another issue in the paper focused on consent from parties to be recorded. The paper also described emerging privacy issues such as devices transmitting and storing data in the cloud and user ability to disable functions or recognize when devices are recording. “Our expectations will evolve more quickly in some areas than others, and so the manufacturers of devices that are introducing microphones for the first time -- like televisions and toys -- should go the extra distance to provide additional transparency and in many cases greater levels of control and choice,” FPF Legal and Policy Fellow Stacey Gray said in a news release.