Alexa is listening to your every word.  So is Siri.  And Cortana.  And Google.  Or the HAL 9000 computer.

Just say the magic words and the devices from Amazon, Apple, Microsoft and Google will obey your commands, give you help with your homework, do math and complete your Christmas shopping (particularly Alexa).

Alexa is a feature on a new device called Amazon Echo, which – unlike the other mentioned devices – is intended to be a stand-alone, Internet connected microphone that sits in your home and listens to everything everyone says.

All the time.  And, when it hears its name “Alexa…” it takes not only what it just heard, but also a few seconds BEFORE it was awakened, and sends the query or comment to Amazon’s cloud servers, which process and stores them.

Want to order Aunt Betty some pecan pie for Thanksgiving, just ask Alexa!  But there’s a price for this convenience.  And not just the $199 for the device (only $99 for Amazon Prime members!)

The price is your immortal soul.

Well, not exactly.  But close.  What all of these devices have in common is that they have always on microphones (or at least the option for “always on”) which listen to and process every word, sound and phrase you utter.

While they are listening for their “wake up” word (Alexa, Hey, Siri, OK Google) they are processing everything.  That means that a clever hacker (or app developer) can use this functionality to listen for other things as well.

The police can look for “key words” or phrases on every one of these devices – as long as they are listening – whether they are on or off.  Marketers can use the device to listen for what shows you are listening to (background sounds) what products you mention, what interests your express.

Then can analyze how many different voices they hear, how many people are in the house, at what times, what dialects or languages or accents they have, etc.  Did I mention that they are always on microphones?  They would know if you are profane or demure.  Loud or quiet.   Everything.  And that’s even before they are “turned on.”

When you invoke the device, you then transmit your comment, question, demand, or request to each of the company’s cloud servers.  That’s not just true for these command level apps.  When you use any voice recognition program (say dictating a memo or email) you are also dictating to the cloud.

So if you say, “Hey Siri, send a text to Jane Doe, Jane, I am just going to finish my drug deal here, but I will meet you in 15 minutes…”  The actual content of the text is covered by the provisions of the Electronic Communications Privacy Act (18 USC 2701) and the Wiretap Law (18 USC 1511).  The government can’t “intercept” that communication while in transmission without a warrant.  Mostly.

But the voice recognition part of the text message may not be so protected.  If I dictate a memo to my lawyers “Hungadunga, Hungadunga, Hungadunga & McCormack..” using any of these voice recognition programs, the contents of my voice is sent over the cloud to the company’s voice recognition engine.  That probably does not constitute a “communication” since I am just talking to a computer. Or maybe it does.  Who knows.

Certainly a clever policeman, prosecutor, judge or hacker could assert that it wasn’t a communication.  Therefore, no warrant would be necessary to “intercept” the contents of my voice.  Once my voice hits the Amazon, Apple, Microsoft or Google voice processor, they turn it into recognizable text.  This is done not just by generalized voice processors, but also by context.

They know the names of my friends and relatives (and the spelling of their names) from my contact and email lists, the names of local stores from maps and GPS, and word choices (there, their, they’re) from context clues.  They are listening.   The voice processing is similarly not a “communication in transmission” and is subject to search and seizure, or hacking.  The voice processor then returns text results to the user.  And it stores the request.

These stored requests, like your browser history, are stored by the cloud provider for as long as they want, although if you are knowledgeable enough you might be able to delete them.  So now the things you do every day and casually are being recorded, processed and stored.

What makes Amazon’s Echo disturbing in the Internet of Things is its ubiquity.  As the IOT becomes more user friendly, it becomes more useful, and therefore more used.  That’s cool.  But as it becomes more used, it collects, stores, and processes more data – and more personal data – about everyone.  Passively.  Without us even knowing it.  If you move the yogurt in the fridge to get to the chocolate pudding, the motion sensor in the fridge knows, and can report you to Weight Watchers or to Chobani.  And it can keep a record of this for posterity.

All of this data will have to be securely stored, deleted, and used properly.  And I doubt we have the ability to do this.  It’s a potential gold mine for marketers, profilers, hackers, stalkers, foreign governments, terrorists, or prosecutors.  It’s subject to misuse, sharing, and subpoena.  If we currently have a wave of data, what this will bring is the kind of wave you see on the water planet in Interstellar.

So before you ask your Amazon Echo to do things for you, ask a better question.  Who else is listening?

Leave a Reply