Contractors Also Listen to Some Siri Audio Without Users’ Explicit Knowledge

With Amazon and Google’s voice assistants confirmed to have humans listening to recordings, it was only a matter of time before it was known whether anyone does the same for Siri.

Alex Hern, the Guardian:

Apple says the data “is used to help Siri and dictation … understand you better and recognise what you say”.

But the company does not explicitly state that that work is undertaken by humans who listen to the pseudonymised recordings.

Apple told the Guardian: “A small portion of Siri requests are analysed to improve Siri and dictation. User requests are not associated with the user’s Apple ID. Siri responses are analysed in secure facilities and all reviewers are under the obligation to adhere to Apple’s strict confidentiality requirements.” The company added that a very small random subset, less than 1% of daily Siri activations, are used for grading, and those used are typically only a few seconds long.

A whistleblower working for the firm, who asked to remain anonymous due to fears over their job, expressed concerns about this lack of disclosure, particularly given the frequency with which accidental activations pick up extremely sensitive personal information.

Hern confirmed to me that their source works for a third party, not Apple directly. That’s similar to the way Google does it, while Bloomberg describes the people who listen to Amazon’s recordings as “employees and contractors”. I think it matters whether these individuals are employed directly or through third parties like Apex — both for the sake of the employees as well as the inherently private nature of what they’re dealing with.

I also think it matters what kind of company is using this data. Amazon and Google use users’ behavioural data to sell targeted advertisements. While they’ve denied that they use voice data for targeting, I still find it slightly more uncomfortable that they may keep records of users’ voice recordings than I do for Apple which doesn’t build huge user data profiles for advertising purposes. In other words, while it’s reasonable to be upset by similar revelations, there are different reasons to be concerned.

Even so, there should surely be a way to opt out entirely and not allow any of your Siri conversations to be selected for review. It’s absurd that there seemingly isn’t a way to do this — turning off Siri entirely is not a solution — though I’ve reached out to confirm if disabling the analytics sharing options in Settings would opt users out. Also, as with Google, I have to question why users are not first asked whether a human can review their audio recording. Less than one percent is a very small proportion, but is still probably a lot of recordings per day given the half-billion devices it’s used on.

It’s pretty strange to me that this is an issue at all for Apple. I’m reminded of their introduction of object recognition in Photos. As Craig Federighi said on the Talk Show in 2016, “if you want to get pictures of mountains, you don’t need to get [them] out of people’s personal photo libraries. We can find some pictures of mountains”. Why should audio recorded in users’ homes, workplaces, schools, and gyms not be treated with a similar or greater level of sensitivity?