An important goal of technology is to make consumers’ lives not just easier but also safer. Technology—in particular
artificial intelligence (AI)/machine learning (ML)—can help to safeguard people’s safety, dignity, and privacy in the
face of various forms of abuse that plague society, including cyberbullying, domestic abuse, and threats to child
safety.
Some of the most heinous crimes we strive to defend against are sexual crimes targeting children and minors, such as
Child Sexual Abuse Material (CSAM). Detecting and stopping the spread of CSAM on the internet by using AI/ML techniques
is a valuable tool that technologists are deploying in various shapes and forms.
To that end, Apple announced recently that it will deploy its NeuralHash technology to detect known images of CSAM. The tool will scan images during the (otherwise encrypted) process of
uploading photos to Apple’s iCloud platform. Apple’s approach will scan for known fingerprints of such material drawn
from the America’s National Center for Missing and Exploited Children database of CSAM.
Apple’s approach will certainly bolster the fight against CSAM by providing valuable information to law enforcement.
Many security researchers, however, have expressed concerns about the privacy implications of the approach. All photo
uploads from all Apple iCloud users in the U.S. will be scanned.
For years, Apple has been subjected to pressure from law enforcement agencies to allow access to encrypted user data,
including photos and messages. As such, security and privacy experts are concerned that these types of “global” scans
may be misused, including by governments who aim at surveilling their citizens. Indeed, due to privacy concerns
expressed by multiple parties, Apple has decided to delay the rollout of this feature.
Many people (especially during the COVID-19 pandemic) use their smartphones to take pictures of private documents or
other types of private data or moments. Given that the potential misuse of content scanning systems raises serious
privacy concerns for the majority of people (who are non-offenders), it is worth considering what can be done to protect
legitimate private information, while still allowing Apple and other vendors to scan for illegal explicit content.
Norton Labs recently released the SafePic app on iOS, which can help with photo-related privacy concerns. SafePic employs AI/ML in order to identify photos on
smartphones containing personal information (such as photos of IDs, credit cards, etc.). It offers users the ability to
protect such photos either by saving them in a secure private vault, or by introducing “smart blur” through the
PhotoBlur feature.
The ability to automatically identify specific classes of sensitive/private photos enables SafePic users to help protect
and make wise decisions about the way such photos are used. SafePic also offers the ability to take photos directly
within the app and add them to the user’s on-device SafePic secure vault. All scanning, “smart” classification and
processing of content happens on the user’s smartphone and none of the sensitive content leaves the device. The SafePic
app is one of the first of its kind and provides consumers with the ability to help safeguard their private information.
You can read the full blog here and you can find SafePic on the Apple app store.