In the tally of privacy grievances, Google’s scanning of emails for child pornography should rank very low, if it appears at all.
In the tally of privacy grievances, Google’s scanning of emails for child pornography should rank very low, if it appears at all.
The controversy bubbled up after law enforcement last month arrested a Houston man for sending images of child pornography from his Gmail account. Google’s automated system detected the content, and the company then forwarded the information to the National Center for Missing and Exploited Children (NCMEC). The arrest sparked a backlash from privacy experts, including an article in The Post titled, “How closely is Google really reading your email?”
In fact, Google — along with Microsoft, Facebook and Twitter, all of which use similar technology — isn’t “reading” your messages. These companies use digital fingerprint technology, pioneered by Microsoft, to automatically identify known child pornography photos that pass through their systems. When there’s a match, a tip is sent to the NCMEC as required by law. No humans sift through attachments, nor do they adjudicate whether certain images qualify as child pornography.
Yet, critics say Google is assuming the role of a “policeman.” While required by law to report images once they’re found, Google wasn’t compelled to install a system that searched for the pictures. Scanning emails for keywords to sell ads, it’s argued, is to be expected — but reporting to authorities the private contents of emails is a step down a slippery slope.
Perhaps theoretically.
According to Google’s terms of service, the company “may review content to determine whether it is illegal or violates our policies.” The blanket statement about illegal activity could allow Google to scan email for activities such as piracy or intellectual property theft. This language opens up the possibility of cases such as one involving Microsoft last year, in which the company searched a blogger’s Hotmail account for alleged trade secret theft.
But that’s very unlikely in practice. The child porn technology was designed solely to match known images from a specific database. It can’t be used to search text for general criminal activity. Those searches would be much more complicated and probably yield a much higher error rate. Google, for its part, views child sexual abuse imagery as unique, morally and legally, and says this is the only situation in which the company hands over user account information to a third party.
Exploitation of children is a heinous crime, and Google is being a “proactive neighborhood watch,” says Meg Ambrose, a Georgetown law professor. That’s an apt comparison.
The technology is narrow and extremely accurate — and a powerful tool in helping eliminate content that has no place in our society.
— Washington Post