Walls have ears: the balance between convenience and privacy

More than five million homes in the United States already have an Amazon Echo virtual assistant, and many more are expected to be purchased over the Christmas season. Now, the device has hit the headlines after police in Bentonville, Arkansas asked Amazon for recordings from a device as part of their investigation into a murder. Few details of the case have been released, other than that the accused is a man who allegedly invited a co-worker to his house who was later found dead in the bathtub.

For the moment, the company has refused to facilitate the recordings, presumably concerned about the idea that Amazon Echo is a kind of spy in your living room. In fact the device only starts recording after a voice command or wake word has been uttered, which in the case of Amazon Echo, is its default name, Alexa — or when it believes that it may have been said. After which it stores those recordings on the cloud, where they can be reviewed by the user or deleted.

The device is permanently alert and listens to us all the time through its seven microphones, but it only records, sends to Amazon and processes those recordings after it has detected its wake word, which in the case of the Echo is Alexa. In reality, very few users access recordings of their voice commands, let alone bother to delete them. In practice, the device is a convenient way of carrying out tasks such as turning on lights or accessing entertainment and content, or of course, ordering goods through Amazon. Few people would see it as a spy or think that recordings of their commands would be of interest to anybody, let alone the police.

Amazon Echo’s role in the Bentonville murder is minor: the device was in the kitchen, the murder allegedly occurred in the bathroom, and the probability that the device has captured anything that might clarify what took place is unlikely: nobody is going to ask their Echo how to commit murder, and the chances that something might have been captured if the device was used at some point to play music are minimal, although this does not prevent the police pursuing that line of enquiry. The chief suspect is a tech fan who lived in a smart home equipped, in addition to an Echo, with other devices such as a Nest thermostat, a Honeywell intelligent alarm, an outdoor weather monitor with wireless transmission, a WeMo light control, a smart meter for water consumption, an iPhone 6S, a MacBook Pro, a PlayStation 4 and four tablets.

Could any of those connected devices help solve the case? It turns out that the water meter has recorded that about 530 liters of water were used between 1 am and 3 am on the night of the incident, leading police to speculate that part of that water, in addition to filling the bath, could have been used to hose away blood.

Should a device as an Amazon Echo be checked as part of a police investigation? Beyond the crime in Bentonville, debate about what data is generated by such devices, what control users have over them, or how they can be used is relevant, and opens up an interesting field about the possibilities of our increasingly connected homes and the right of the authorities to analyze the constant flow of data we generate. Should Amazon alert the police, for example, if their device captures certain recordings, either purposely or accidentally, that suggests a crime has been committed? Does the smart home mean we need to rethink the pay off of so much convenience?


(En español, aquí)