When people use Apple's Siri and Dictation, the Cupertino giant collects that data and stores it on its servers for two years before discarding it.

Apple collects data to improve its voice-controlled Siri service, but it reportedly makes that data anonymous. The company detailed its practices after one civil rights liberties group criticized Apple for not doing enough to inform its customers on their privacy rights. The privacy advocates called on Apple to disclose exactly what information it collects and stores about users, and the Cupertino giant delivered.

Company spokeswoman Trudy Muller told Wired that Apple collects anonymized data with the sole purpose of improving its service, as the company takes user privacy very seriously. As users mostly operate Siri remotely, the personal assistant software requires a data connection to operate. The popular software is available on iPhone, iPad and iPod touch.

Apple stores voice clips and assigns random numbers to each clip to represent the user who recorded it. Apple does not associate the number in question with an Apple ID, email address or anything else that could personally identify the user.

The random number ceases to represent the saved clip after six months, but the audio file may remain stored on the company's servers for up to two years for "testing and product improvement purposes." If a user decides to turn off Siri altogether on their device, however, Apple will delete the randomized number and any associated data.

"Dictation is turned on as part of Siri," reads Apple's support documentation. Both Siri and Dictation fall under the same service-level agreement, therefore have the same privacy policy. If the user keeps Siri and Dictation on and not disabled, however, Apple will keep any voice-input data and corresponding personal information for up to two years from the time it was first entered into the device.

Security advocates and various companies have long criticized Apple for collecting Siri data before the service provides results to users. Last year, for instance, IBM banned the use of Siri on its corporate networks specifically for this reason, fearing that sensitive information could leak through Apple's virtual assistant.