Apple keeps your conversations with Siri for two years

Apple today announced that data from conversations with Siri and voice dictation are kept on their servers for two years. The announcement arrives after one civil liberties group suggested the Californian company aren't active enough in informing customers of privacy rights.

Exactly what Apple's voice assistant uses the collected data for isn't clear, but the company did confirm that users' questions can be kept for up to two years. Siri, which requires a constant Internet connection to function, sends queries directly to Apple's servers before users receive a response from the recognisable robotic voice.

Whenever you speak to Siri, the data is shipped off to Apple's data farm for analysis. The company generates random numbers which represent the user's device from which the data was sent from, and associates all subsequent audio files with the generated number. After six months, your user number is "disassociated" from the stored file, however the disassociated files are kept on Apple's servers for up to a further eighteen months.

Apple's Trudy Miller said in an interview:

Apple may keep anonymized Siri data for up to two years. If a user turns Siri off, both identifiers are deleted immediately along with any associated data.

Despite Siri data not being directly linked to users and only identifiable by a randomly-generated number, but Nicole Ozer suggests people should be careful about what things they're saying to their personal assistant:

Transcripts of what you say to Siri could reveal sensitive things about you, your family, or business. Siri works for Apple, so make a note to yourself to really think before you speak.

While Ozer may be right in saying that ultimately your spoken data ends up in the hands of Apple, the associated privacy issues are so small they're barely worth mentioning. Collected data is used by the company for the purposes of improving user experience and to create a better product in the future. Despite this, Ozer is right in saying Apple need to do more to make the public aware their data was being sent over to Apple after inputting it into their own device.

Source: ZDNet

Report a problem with article
Previous Story

Bill Gates supporting new technology for vaccines

Next Story

Twitter ending Internet Explorer 6 & 7 TFW support

16 Comments

Commenting is disabled on this article.

Discussions is a bit of a push. Surely 'interactions' would be a better term

In all seriousness, i kind of expected it to keep a lot of data and send it out to Apple. Voice recognition stuff really needs to, to become more and more effective.

Wish Siti saving the data for 6 months at a time, I have not seen her use that to improve her recognition of my British accent. I have to talk like a Yank to make it understand me. Even when I correct her, she still does not learn.

If your Siri is a "she", you don't have it set to British English. British English Siri is actually a "he". The USA Siri is a "she".

iKenndac said,
If your Siri is a "she", you don't have it set to British English. British English Siri is actually a "he". The USA Siri is a "she".

Always found that odd. An age old British survey discovered that we react in a better way to a female voice, hence all the public transport/tube announcements being recorded in a woman's voice.

So you set your Siri to USA English and are surprised that it understands and American accent better than a British one?

Incredible.

No, don't be a smart ass. Siri is mean to learn your voice.

So if you pronounce "RED" as "BLUE" and you correct Siri, each time you say "Blue", it should recognize it as "Red" for your Apple Account. This feature fails and so it does not learn your accent.

i understand saving it to use to make it better, but if i say "where is the store?" how does it know what to use that for unless i tell it what i'm saying ahead of time? how does it know what i've said? and if it knows, why save it?

Buttus said,
i understand saving it to use to make it better, but if i say "where is the store?" how does it know what to use that for unless i tell it what i'm saying ahead of time? how does it know what i've said? and if it knows, why save it?

Because it's excellent training data for improving algorithms, including improving recognition in environments with lots of background noise.

Yes, they might know you said "where is the store?", but that knowledge is necessarily. That's how supervised learning works.