I'm trying to decide whether I care about this or not. My bank records our telephone conversations, and those are much more detailed than an Alexa or Siri request, and are directly associated with my personal information. This seems pretty benign.
The only real risk to me is embarrassment, but it's not likely someone I know personally is ever going to hear me saying or doing something I'd wish they hadn't. It's true that I don't want an Amazon contractor hearing my passwords or financial codes, but in the absence of a way for them to determine exactly who I am, even that information is essentially useless to them.
Yes, but in my country everyone who is recording the call needs to inform you and ask everytime you speak with them for your consent. Otherwise the call cannot continue
Okay. So what? Once you give permission, that's what happens. If you refuse, you don't get your banking done. I don't understand your point.
Not super-germane to the article, but this isn't strictly true. You can always (nicely) ask the operator on the other end to not record the call, and guess what, it works sometimes. You may need to bump up a level, because it's not a request first-level support people get every day, and you may need to be persistent, but (in most cases) there are no laws that require companies to record conversations, it's merely what happens by default. I'm sure it varies highly by company, and of course YMMV.
Nothing to see (or hear) here. If Apple isn't doing this, well at least that would explain why Siri doesn't seem to improve as fast as other virtual digital assistants.
Not a surprise. Do you remember when Alexa sent conversations recorded to private company? Siri is different. Anonymous phrases random can be analysed to improve it but all the requests still on the device , only numbers code send to the servers . Completely different . Amazon and google make money with personal informations and conversation . To do that they need a profile so thy read the conversations connected with name and surname.
What people don’t realize is that so much of AI is a Potemkin Village of behind the scenes human intervention, especially AI chat bots are usually humans.
Those who criticize Siri for an inability to have a fluid conversation like Alexa? Well this is the cost. The choice is up to you.
Personally I like telling Siri to do simple tasks for me...set a timer, what’s the weather, dim the lights, etc. I rather not have a conversation and leave my privacy intact, but that is my choice.
The lack of transparency and anonymity is troubling. The article mentioned “banking information” but I wonder how many people are actually using explicit information about their account to the point it could be compromised. My guess is zero to not very many. I mean, when was the last time you came home and said, “Hey, honey! I just wanted to let you know that I deposited that $10,000 into our Bank of America checking account # 520439203949!”?
Making a food delivery order without an app. I have said my debit # out loud hundreds of times. You don't have to be directly talking to the speaker for it to be listening..right? I don't own one so I am really not 100% sure..
Some folks give bank account info over the phone to pay a bill as well. I don't but their are people that still handle the old fashioned way, that could easily own one of these devices.
So I agree that most people don't walk in and say hey honey just wanted to let you know.... but there are scenarios where people give sensitive info out loud while near these devices.
I'm trying to decide whether I care about this or not. My bank records our telephone conversations, and those are much more detailed than an Alexa or Siri request, and are directly associated with my personal information. This seems pretty benign.
The only real risk to me is embarrassment, but it's not likely someone I know personally is ever going to hear me saying or doing something I'd wish they hadn't. It's true that I don't want an Amazon contractor hearing my passwords or financial codes, but in the absence of a way for them to determine exactly who I am, even that information is essentially useless to them.
Yes, but in my country everyone who is recording the call needs to inform you and ask everytime you speak with them for your consent. Otherwise the call cannot continue
Okay. So what? Once you give permission, that's what happens. If you refuse, you don't get your banking done. I don't understand your point.
They just need to make the person aware the call is being recorded. You don't give permission for anything. I don't get the OP's point either?
Amazon is doing what Apple should be doing - working to make the voice recognition the best it can be. As long as there are proper safeguards, and it seems there is, the only way to really improve is to have people listen to some audio and compare it to what was transcribed and make adjustments if necessary.
This part of the BBC article is more telling than any other part of the article.
“Can I stop human reviewers listening to my voice clips?”
even though Google and Amazon claim to anonymize your recordings, they have a feature that you can review your recordings and delete them. If it is anonymous, how do they know it is your recording.
Apple gave the best answer in that they cannot get your recordings because they don’t know whose recordings they have.
Giving my information over the phone vs. my private conversations at home being “sampled at home” are two entirely different things. I don’t even see why that is close to a comparison. Government isn’t suppose to intrude on phone calls without a warrant but we put our trust in companies. I put my trust in a company that isn’t making money on my information or purchasing habits. Follow the money.
I'll stick with HomePod thank you very much! Worth every penny!
Are commands processed completely on-device? I thought any commands were sent back to the mothership for processing into device directives.
If they are not processed on-device I would expect that Apple has a similar group to Amazon that is tasked at reviewing recordings to improve recognition.
I'm trying to decide whether I care about this or not. My bank records our telephone conversations, and those are much more detailed than an Alexa or Siri request, and are directly associated with my personal information. This seems pretty benign.
Um, yeah, you should. Your bank records calls that you initiated, just as Apple saves Siri requests you made.
Neither are always listening in the background. Well technically, iOS devices are listening in the background for "Hey, Siri", but that's done at a hardware level and not recorded.
This is why I refuse to buy anything that's marked"Alexa built-in!" That means there's a microphone in it. I much prefer HomeKit... where the appliance itself doesn't listen, but relies on an iOS device to send it commands and control it.
So the only difference between what Amazon and Apple are doing is that Amazon is keeping an identifier token in case they need it, but in both cases the individual people doing the listening have no way to identify the person?
Maybe I'm missing something, but I'm totally fine with the way Amazon's doing it. Apple's way is better for privacy, but Amazon's way is likely better for improving the service. I expect Amazon wants to keep the identifier so that can isolate people with accents/weird phrasing/etc or to see how updates affect people who previously had trouble with a given request.
Edit: I use the Echo in my house because it works so much better for me for price, features, integrations, and options. However, if Apple had the better product (for me), I'd buy it because I do value privacy. I just don't feel like Amazon is doing anything wrong. That said, I would never, ever, ever let Facebook devices in my house.
And how do you know Apple isn’t recording and maybe selling all your data? Oh, wait... you don’t.
I trust Apple more than the others because it makes money by selling hardware and services, not by selling their customers. And because they have a good track record on privacy.
The report notes Amazon's review system strips identifying information like a user's full name and address from the clips, but leaves the customer's first name and product serial number intact.
Apple, too, employs a human review process to improve Siri. In a security white paper (PDF link), Apple notes Siri saves voice recordings "so that the recognition system can utilize them to better understand the user's voice." The recordings are stripped of identifiable information, assigned a random device identifier and saved for six months, over which time the system can tap into the data for learning purposes. Following the six-month period, the identifier is erased and the clip is saved "for use by Apple in improving and developing Siri for up to two years."
This would make an excellent test question for whether company A and/or company B is correctly complying with EU directives with respect to private personal data.
First name + serial number undoubtedly fails the deidentification test. A random ID assigned? We have a winner!
redraider11 said: If “hate” speech becomes illegal it’s just a matter of time until someone at Amazon, or the AI, takes a joke out of context and the police are knocking on your door.
You can call the police on neighbors if you hear what sounds like violence. What did the Alexa department do with that recording of sexual assault? Sounds like they just passed it around the office with a “not our job” attitude.
So the only difference between what Amazon and Apple are doing is that Amazon is keeping an identifier token in case they need it, but in both cases the individual people doing the listening have no way to identify the person?
Maybe I'm missing something, but I'm totally fine with the way Amazon's doing it. Apple's way is better for privacy, but Amazon's way is likely better for improving the service. I expect Amazon wants to keep the identifier so that can isolate people with accents/weird phrasing/etc or to see how updates affect people who previously had trouble with a given request.
Edit: I use the Echo in my house because it works so much better for me for price, features, integrations, and options. However, if Apple had the better product (for me), I'd buy it because I do value privacy. I just don't feel like Amazon is doing anything wrong. That said, I would never, ever, ever let Facebook devices in my house.
The difference is that Amazon is choosing (by intent or by negligence) to be working with identifiable personal data; Apple choose to work with anonymous data.
The article mentions the position that this puts Amazon in. An employee (or contractor) identifies something that sounds like sexual assault and sends it up the food chain. Management decides not to do anything about it. Amazon as a corporation is putting itself in the position of deciding of how to act when it learns private information.
Imagine the following scenario: Audio from Device SN: 123-123-123-43344. User First Name: James. "[foreground voice]: hey Alexa, order 3 more pressure cookers and 3 more bags of nails" "[background voice:] how many people you think we'll be able to kill? Will Senator so-and-so be there?"
If I heard that as an Amazon flunky, I would call my supervisor. Presumably there would be a great temptation for Amazon to let the police know that James [Look up last name] who registered Echo device 123-123-123-43344 may be committed or has committed a terrorist act.
Apple, on the other hand, could do nothing but let the police know what they heard and hand the audio over for voice analysis.
In this case, theoretically the Amazon approach could save lives. But the Apple approach preserves privacy. You can prefer either one, but they aren't the same.
Comments
Those who criticize Siri for an inability to have a fluid conversation like Alexa? Well this is the cost. The choice is up to you.
Personally I like telling Siri to do simple tasks for me...set a timer, what’s the weather, dim the lights, etc. I rather not have a conversation and leave my privacy intact, but that is my choice.
“Can I stop human reviewers listening to my voice clips?”
Apple gave the best answer in that they cannot get your recordings because they don’t know whose recordings they have.
Giving my information over the phone vs. my private conversations at home being “sampled at home” are two entirely different things. I don’t even see why that is close to a comparison. Government isn’t suppose to intrude on phone calls without a warrant but we put our trust in companies. I put my trust in a company that isn’t making money on my information or purchasing habits. Follow the money.
If they are not processed on-device I would expect that Apple has a similar group to Amazon that is tasked at reviewing recordings to improve recognition.
Um, yeah, you should. Your bank records calls that you initiated, just as Apple saves Siri requests you made.
Neither are always listening in the background. Well technically, iOS devices are listening in the background for "Hey, Siri", but that's done at a hardware level and not recorded.
This is why I refuse to buy anything that's marked"Alexa built-in!" That means there's a microphone in it. I much prefer HomeKit... where the appliance itself doesn't listen, but relies on an iOS device to send it commands and control it.
https://en.m.wikipedia.org/wiki/Amazon_Mechanical_Turk
Maybe I'm missing something, but I'm totally fine with the way Amazon's doing it. Apple's way is better for privacy, but Amazon's way is likely better for improving the service. I expect Amazon wants to keep the identifier so that can isolate people with accents/weird phrasing/etc or to see how updates affect people who previously had trouble with a given request.
Edit: I use the Echo in my house because it works so much better for me for price, features, integrations, and options. However, if Apple had the better product (for me), I'd buy it because I do value privacy. I just don't feel like Amazon is doing anything wrong. That said, I would never, ever, ever let Facebook devices in my house.
First name + serial number undoubtedly fails the deidentification test. A random ID assigned? We have a winner!
The article mentions the position that this puts Amazon in. An employee (or contractor) identifies something that sounds like sexual assault and sends it up the food chain. Management decides not to do anything about it. Amazon as a corporation is putting itself in the position of deciding of how to act when it learns private information.
Imagine the following scenario:
Audio from Device SN: 123-123-123-43344. User First Name: James.
"[foreground voice]: hey Alexa, order 3 more pressure cookers and 3 more bags of nails"
"[background voice:] how many people you think we'll be able to kill? Will Senator so-and-so be there?"
If I heard that as an Amazon flunky, I would call my supervisor. Presumably there would be a great temptation for Amazon to let the police know that James [Look up last name] who registered Echo device 123-123-123-43344 may be committed or has committed a terrorist act.
Apple, on the other hand, could do nothing but let the police know what they heard and hand the audio over for voice analysis.
In this case, theoretically the Amazon approach could save lives. But the Apple approach preserves privacy. You can prefer either one, but they aren't the same.