Apple says its announcement of automated tools to detect child sexual abuse on the iPhone and iPad was “jumbled pretty badly”. On 5 August, the company revealed new image detection software that can alert Apple if known illegal images are uploaded to its iCloud storage.
Privacy groups criticised the news, with some saying Apple had created a security backdoor in its software. The company says its announcement had been widely “misunderstood”.
“We wish that this had come out a little more clearly for everyone,” said Apple software chief Craig Federighi, in an interview with the Wall Street Journal.
He said that – in hindsight – introducing two features at the same time was “a recipe for this kind of confusion”. Apple announced two new tools designed to protect children. They will be deployed in the US first.
The first tool can identify known child sex abuse material (CSAM) when a user uploads photos to iCloud storage. The US National Center for Missing and Exploited Children (NCMEC) maintains a database of known illegal child abuse images. It stores them as hashes – a digital “fingerprint” of the illegal material.
Cloud service providers such as Facebook, Google and Microsoft, already check images against these hashes to make sure people are not sharing CSAM.
Apple decided to implement a similar process, but said it would do the image-matching on a user’s iPhone or iPad, before it was uploaded to iCloud. Mr Federighi said the iPhone would not be checking for things such as photos of your children in the bath, or looking for pornography.
The system could only match “exact fingerprints” of specific known child sexual abuse images, he said.
If a user tries to upload several images that match child abuse fingerprints, their account will be flagged to Apple so the specific images can be reviewed.
Mr Federighi said a user would have to upload in the region of 30 matching images before this feature would be triggered. In addition to the iCloud tool, Apple also announced a parental control that users could activate on their children’s accounts.
If activated, the system would check photographs sent by – or to – the child over Apple’s iMessage app.
If the machine learning system judged that a photo contained nudity, it would obscure the photo and warn the child. Parents can also choose to receive an alert if the child chooses to view the photo. [BBC]
