Some 40% of Kiwis who own a smart device with listening capability say they've received social media posts and ads about a topic they talked about aloud.
That's according to a weighted survey of 1000 adult New Zealanders, release by Unisys this morning.
The survey also found half of respondents (47%) support the government collecting this information to identify who is in the vicinity of a disaster. Yet only 20% support the government monitoring an individual's travel patterns to plan roads and public infrastructure. Nearly four in 10 (38%) support airports and airlines collecting information to efficiently guide a passenger's journey through an airport, but only 10% support an employer doing the same to monitor an employee's location during the workday.
With smart devices, 12 per cent of respondents said a virtual assistant on their smartphone/watch asked for more infomation even though I had not turned the assistant on.
And the same percentage reported a smart speaker had woken up and asked for more information, without them using an activation phrase.
Are New Zealanders right to be paranoid about the likes of Amazon's Alexa, Apple's Siri Google's Assistant?
All three companies have, in the past, copped to recording what you say - but they have also emphasised that that recordings are anonymised, in an automated process controlled by machines, and that their employees can't access individual accounts.
More disturbingly, it has also been revealed that all three - plus Facebook - have given humans access to recordings to transcribe. The spin is that this will help deliver better service.
Just yesterday, the Herald reported that Apple has apologised for allowing contractors to listen to audio recordings of customers' Siri communications in order to grade them.
Last month the Guardian revealed contractors were tasked with grading the accuracy of the digital assistant which had overheard conversations - topics included talk about doctors appointments, drug deals and even couples having sex.
Apple said it had decided to make changes to Siri as a result of a review into the grading programme, it said in an unsigned statement to its website.
"As a result of our review, we realise we have not been fully living up to our high ideals, and for that we apologise," Apple said.
All of the players in this market are now offering tighter privacy controls over how a smart device operates. Privacy Commissioner John Edwards has welcomed that trend, but also said that he strongly prefers a "privacy by design" approach where such controls are obvious and up-front, not hidden in fine print or after screed of other options.
Edwards recently posted a link to Wired's guide to taking charge. Here are some highlights:
When Alexa and Google Assistant capture your voice, they store those recordings indefinitely, but they do provide a way to delete them if you want. Apple, meanwhile, says it analyzes and deletes these snippets automatically, so there's no history for you to comb through. As with Amazon and Google, though, Apple does use human reviewers to evaluate a subset of recordings. These each receive a random ID number and are stored for six months, after which they are stripped of that ID and may be stored for up to two more years.
To delete your Alexa history, open the Alexa app on your phone and go to Settings > History. In this view you can only delete entries one by one. To delete en masse, go to Alexa Privacy Settings on Amazon's website and then choose Review Voice History.
For Google Assistant, go to myactivity.google.com and click the three dots in the upper-right corner. Then choose Delete Activity. Then you can choose a date range - today, yesterday, last 7 days, last 30 days, all time, or custom - for the entries you want to delete. Then navigate through All Products > Voice & Audio > Delete.
In Amazon's Alexa app, open the menu button at the top of the screen. Go to Alexa Account then Alexa Privacy and tap Manage how your data improves Alexa. Then turn off Help Develop New Features and Use Messages to Improve Transcriptions.
Smart speakers are designed to begin recording and processing what you're saying only when they hear their "wake word," but in practice they can often misinterpret other sounds and go a little rogue. One way to know for sure when your Echo or Google Home is recording? Turn on audible alerts so that the device emits a sound every time it spins up. Otherwise, you have to hope you catch the indicator light flashing at the right time.
In the Alexa app go to Settings and pick a device. Then choose Sounds and go to the Request Sounds section. Then turn on start and end of request sounds.
For Google Home, open the Google Home app. Choose the device you want to manage. Tap Settings and then Accessibility under Device Info. Google offers two options: Play Start Sound and Play End Sound. Turn them both on to get the best idea of when a recording session begins and ends.
For HomePod, open the Home app, select your device and then Details. Turn on Sound when using Siri.
The easiest way to ensure that no one is listening through your smart assistant's microphone is to mute the device. Amazon Echo has a mute button on top, and Google Home has one on the back towards the top of the device. Apple HomePod doesn't have a physical mute button, but there is a toggle in the Home app under Details for the device you want to mute. Turn Hey Siri off to cut the mic. You can also say, Hey Siri, stop listening, and then confirm, Yes.
Another way to make sure that no one can remotely access or manipulate your smart assistant is to lock down your Apple ID, Google, and Amazon accounts. Choose a strong, unique password and turn on two-factor authentication to make it harder for anyone to get in from afar.
And don't forget that all three devices also have an additional, almost magical feature that can be used to comprehensively address every type of privacy concern: the power cord.