GMail doesn’t get the raw take of every conversation within earshot, 24/7/365.25, that simple. Any other messaging service shares the same caveat.
Make one named HAL with a red throbbing light. I don’t know how you’d do the voice. I guess you could just record “I’m sorry Dave. I’m afraid I can’t do that.” and play it back for every response.
Exactly. It’s very silly to be paranoid around them.
Even so, I am going to stop walking around my house reciting my credit card number all day…
I agree, privacy is valuable. I suppose my point is I’m not in fact losing any of that valuable privacy, because there is no reason --none— to believe anyone is going to actually spy on me with this device. If I thought they would–no matter how banal–I would of course be upset. But they won’t. So I’m not.
Had to like just for the .25. Be still my detail-oriented heart.
You miss the point. Yes, there IS a difference between the two paradigms,
although both certainly risk your info.
I feel that this post is undermined by your colleagues hawking these things with affiliate links but… meh, we seem to have normalised surveillance a long time ago.
If they ever criminalize following the cat around saying its name over and over because no one else is home and I need to annoy someone, then Bezos has enough to put me away for a very long time.
Otherwise all they have is me saying “Alexa, play Touch My Tooter,” and then my son laughing and my wife telling me to knock it off. (Meanwhile, all her Spotify friends just think she really likes Ween.)
Well… at least when they’re not trying to sell it to you in the glorious BB partner store posts!
i would quibble with “crazy” but willingly accept “highly idiosyncratic.”
Mozilla has just released a new open source Automatic Speech Recognition engine that doesn’t seem to need cloud resources to work.
I don’t think that it’s ready to run native on phone or Echo level hardware, but that’s in the plans.
Mozilla further notes that the plan for the future is to release a model that’s light and fast enough to run on a smartphone or single-board computer like the Raspberry Pi.
Now, a lot of stuff will require accessing the Internet, “What’s today’s weather?”, but the actual microphone data won’t be on the Internet, and local commands to turn lights on or adjust the thermostat won’t be either.
(Whinging from law enforcement and spies that “this makes our job harder!” in 3… 2… 1…)
not so much paranoia as it is a sense of unease at having my location, my voice messages, my credit card/debit card data, etc. condensed and accessible to the point of being broadcast in ways uncontrollable and very difficult to track.
as for the microphone, i physically disable any built in mics and use an auxiliary which can be removed when done.
Today at the staff meeting, after another discussion on the latest security-theater measures that we have to endure, my boss said that higher-level managers were thinking about how to use this kind of devices at our workplace. I was like, sure, having a listening device manufactured in China should be great way to comply with ITAR regulations.
I thought I remembered reading that while they may have exabytes of storage, they ultimately have far too much useless data to sort through. But I’m not sure where I saw that exactly.
I’m sure that companies like Palantir Technologies will offer to jump into that gap.
Yep, nothing wrong with beltway parasites selling cybercrud, with alt-right aligned owners, having access to questionably gathered surveillance intelligence.
First off, the device listens locally for the trigger phrase. There is no sending of all live recordings over the internet — that would be insane for reasons of, well… physics?
I like hearing what people actually do with these things because I find them pretty darn useless in practice. Privacy is not the problem, the problem is these devices just don’t do anything particularly useful, outside of some very niche needs.
The Echo isn’t always sending recordings back to Amazon. All security teardowns I’ve seen agree with Amazon’s assertion that it only sends sound recordings after it is woken up with a trigger word.
Cory’s post is sadly missing the most relevant fact from the very murder investigation that he referenced. The police were unable to retrieve any data from Amazon, because the Echo products don’t send audio data to Amazon’s servers unless the wake word is spoken. It doesn’t matter how willing Amazon was to comply, they simply don’t have it.
Irony was still perfectly willing to lend a hand, however. His Echo Dot may not have proven to be a reliable IoT witness for the prosecution, but evidence was collected from his smart water meter. He allegedly murdered his victim about 1:00 AM or so, and around 3:00 AM, his water meter showed the anomalous use of a hundred gallons of hot water. The investigators claimed he used the water washing up the crime scene.
Moral of the story: if you’re about to be murdered, no matter where you are, loudly say “Alexa, Siri, Hey Google, Cortana*, I am being murdered!” Some nearby phone or audio device might pick it up.
* Yes, that’s a joke. Nobody has Microsoft phones.
You wake it up with an “OK Google” or whatever, but what ends the stream of data being sent? How long a period of silence is required? If the parsing is done remotely, then presumably the message to stop streaming is sent back from the server, not determined locally.