Apple’s Siri voice assistant on iOS devices retains information to help the company generate better and more accurate results in the analysis process that takes place at its remote servers. The company has never previously revealed exactly how long it keeps that data or how exactly it works, but now Wired has learned from an Apple spokesperson exactly how Siri IDs and store data, as well as for how long.
Apple said that it first associates voice files gathered by Siri with a randomly generated number to keep all associated data anonymous, and separate from any other identifiers including your Apple ID or even your email address. That number remains tied to data for six months, at which time Apple deletes it, while retaining the voice file. The voice file itself, now separated from any kind of identifier, can live on Apple’s servers for up to 18 more months in order to help Apple refine Siri and test new products. When a user turns Siri completely off, however, all data and identifiers are immediately deleted.
Privacy concerns around voice dictation services are nothing new. Nuance, the company that helps power Siri with its voice-recognition software, had to defend against privacy concerns back in 2009 when it launched Dragon Dictation for the iPhone. It, too, stores transcriptions of conversations on its servers to help improve its own technology’s results. The situation isn’t all that different from the type of information Google collects to make sure that its ad targeting works effectively, or to help services like Google Now operate properly.
The main concern of privacy critics like American Civil Liberties Union lawyer Nicole Ozer, who sparked the Wired investigation to begin with, seems to be around the fact that Apple doesn’t include this information about its data retention policies anywhere that’s easily accessible to users of Siri, like in its FAQ page about how Siri works. Her argument is built around Apple’s duty to keep consumers informed, since it could influence what type of information they share with services like Siri.
The bottom line is that if an app or service requires a data connection, in all likelihood there’s a back and forth transmission of information going on, and if privacy is one of your top-of-mind concerns, you should be cautious in any such situation. Apple’s policies with Siri seem no more or less egregious than any other, but it is nice to see the company spell it out in no uncertain terms.