CBI ASKS SOCIAL MEDIA TO USE PHOTODNA TO TRACK SUSPECTS IN INDIA


Looks like surveillance is going to be the big theme going forward in 2019.

It’s not even been many days since the MHA notification which gives 10 government agencies the rights to snoop on you, provided they get the right permissions from a competent authority, there is already another request being made by the Central Bureau of Investigations.


According to a report in The Indian Express, CBI has asked social media platforms to start using PhotoDNA for investigating regular criminal cases. In a notice issued by CBI under Section 91 of the CrPC, it states that social media platforms must run PhotoDNA on images of suspects who are being investigated. “The said information is required very urgently for the purpose of the investigation,” said the notification.

What this essentially means is that social media platforms have to run the PhotoDNA surveillance search on all the user images there on its servers, as opposed to doing this for the photo of the suspects in question. What is essentially a tool meant to be used for containing child exploitation is being employed for general investigations.

The only issue is that this request is in violation of international norms which mandate that this technology can only be used to identify child exploitation images. Microsoft has given the software for free to law enforcement agencies, via the forensic tool developers, around the world to ensure child exploitation or child pornography is eliminated using this tech. It is not meant to be used for any other purpose, as it will then legitimise mass surveillance and impose restrictions on free and open internet.

Asking social media platforms to run PhotoDNA on their servers is also in violation of the right to privacy which has been upheld by the Supreme Court. If PhotoDNA is run on its servers by social media platforms, it places everyone under surveillance irrespective of whether one is a suspect or not.

PhotoDNA: All you need to know

PhotoDNA is a free-to-use Microsoft-owned product which assists with photo authentication. It is not the same as a facial recognition software. Images that you may come across on the web, for instance, they may have been either uploaded from the source or reused or altered and shared. PhotoDNA helps in creating a unique signature for a photograph using certain parameters, which makes it easily identifiable.

In essence, it first converts a photo into black and white, resizes it and then divides the photo into various grids. It creates a hash or a unique signature of the photo, from the histogram data (gradient information) which is unique to that photo. This DNA of the photo will remain the same even if the photo has been edited. When you run this PhotoDNA through an image database, any image which matches with DNA can be identified quickly. A PhotoDNA hash cannot  be used to recreate an image however.

The idea behind PhotoDNA is to be able to pick out images related to child exploitation which may have been shared across the internet. With one photo or video, PhotoDNA can extract the unique signature of that image and then reference it against the image database where the photo or video may have been shared or replicated online. Copies of the image can be identified and law enforcement can ensure that illegal content — child pornography — is taken down by the websites hosting them. Doing this manually or by any other photo matching technique can take quite a while, but PhotoDNA claims that its technology significantly reduces that time frame.

“Microsoft donated PhotoDNA to the National Center for Missing & Exploited Children (NCMEC). NCMEC is the United States clearinghouse and comprehensive reporting centre for all issues related to the prevention of, and recovery from, child victimisation, including abduction, abuse and exploitation,” according to the PhotoDNA page.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

error: Unauthorized Content Copy Is Not Allowed
%d bloggers like this: