Larry Magid: How Google Watches Users’ Gmail Accounts
SAN FRANCISCO (KCBS) — A 41-year-old man suspected of distributing child porn was arrested in Houston after Google tipped off the National Center for Missing and Exploiting Children after flagging photos in the his email account.
John Henry Skillern, a registered sex offender convicted of sexually assaulting an 8-year-old boy in 1994, was detected by Google allegedly sending explicit images of a young girl in an email to a friend.
While it’s a good thing that they are able to help identify people who may have child porn, what does it say about our own privacy online? Just how much do they know about the average user?
Firstly, I should probably mention that I am one of the board of the directors for the NCMEC and through my nonprofit ConnectSafely.org. I do work with Google, so I’m quite familiar with this issue.
There is a federal law to requires any internet service provider to report any images of child pornography to the NCMEC, who then analyses them and if it’s determined that a law has been broken, they will refer it to federal or state authorities. So Google, of course, has to do this by law.
Google uses a licensed technology called Photo DNA which was created by Microsoft. It essentially analyzes every known child porn image and then it looks to compare them for matches. So if someone sends a child porn image through Gmail, it will scan it, recognize it and report it. That doesn’t mean they know what you’re writing about but if a child porn image is sent, it will be scanned and reported; it’s federal law.
It’s well known that Google scans users’ email and presents ads based on that. If I send an email to someone about Hawaii, I might receive ads for Hawaiian trips, but that’s a machine that does that—not a human being.
Google, Facebook, Yahoo! and other services all use this hashing technique. So this case in Houston is sort of unique but it’s unrelated to other forms of law enforcement.