The employee uploaded explicit pictures in cloud account

Oct 31, 2016 07:16 GMT  ·  By

A Microsoft employee was arrested after police investigators found child pornography in his cloud storage account thanks to a tip sent by the Redmond-based software giant itself.

Gregory Rentfro, 49, of Duvall, who worked for Microsoft as a security engineer, uploaded four explicit pictures of minors involved in sexual acts. Redmond itself alerted the authorities, but it’s not yet clear if the company knew that it was its own employee or not.

“One of their customers had uploaded four images of minors engaged in sexually explicit conduct,” the court documents reveal according to KIRO7. “Rentfro admitted to finding depictions of minors engaged in sexually explicit conduct from the Internet. He said that he’d been saving them to an external hard drive until Microsoft closed his OneDrive account.”

Microsoft has already issued a statement to explain that it’s up to the authorities to conduct the investigation here, explaining that “the employee is on leave” until a ruling is given.

At this point, Rentfro is held in the King County Jail on a $75,000 bond, as investigators believe that he poses a risk to his own child.

How Microsoft detects child pornography pictures

When hearing such news, the first thing that comes to many people’s mind concerns user privacy, as Microsoft can be easily accused of looking into accounts for photos that could violate its guidelines.

But in fact, Microsoft moves no finger, and the whole job is performed by a technology called Photo DNA, which automatically scans accounts for photos that involve online child sexual abuse.

Furthermore, Photo DNA can help remove child pornography from the Internet by scanning for hashes of pictures that have already been flagged as illegal. “The program protects user privacy in that it doesn’t look at images or scan photos; it simply matches a numerical hash against a database of known illegal images,” Microsoft explains.

Work on Photo DNA first started in 2009, and the first public implementation was released in 2011, with the service now based in the cloud with a growing database of photos that are labeled as illegal.