Goodbye Photoshopped images: Izitru launches to verify original photos

It’s not uncommon to see an image go viral — then get called out for being fake. News organizations unknowingly published false images of Osama bin Laden after his death, and there’s always the rounds of fake viral photos that are shared on Facebook and Twitter.

To combat photo falsification, FourandSix technologies introduced izitru, a free photo verification app and website that uses six forensic tests to determine a file’s authenticity. It’s not just looking at the metadata, said FourandSix co-founder Kevin Connor. The tests look at everything from how the file was compressed to what camera was used to whether the file has been resaved and manipulated.

“Every device and every piece of software has a somewhat different way of storing their JPEG files,” Connor said. “The most significant way that things differ is in the compression settings that the devices use.”

Image verification is often done by checking the metadata and scanning the image for any visual manipulation — wrong shadows or obvious Photoshop work. A large part of it is just having a healthy skepticism of images and trying to track down the original source. That’s where Connor hopes izitru can step in.

“We were initially addressing this whole issue from the standpoint of the recipient of the photo — what tools can we put in their hands so we they can analyze photos and decide whether they can trust them or not,” Connor said. “We started thinking maybe we need to turn this around and, instead of focusing on the recipient, focus on addressing the images at the source before they’re shared or as they’re getting shared.”

A photographer can upload an image, either anonymously or from their social network, and choose its copyright settings. Izitru will automatically run all six forensic tests and issue it a trust certification. If someone is at the scene of a disaster or even a beautiful sunrise, they could upload the photo to izitru and then distribute the link to journalism outlets or share on Twitter to guarantee its authenticity.

“It’s only going to be the original file as captured by the camera that’s going to pass,” Connor said. “It’s about creating a basic level of trust at the start.”

Given the challenge, I tried my best to fool the app into giving me high trust ratings.

I took a screenshot of a photo of my dog and tried to upload it, but I was rejected because the file was a PNG. I switched it to a JPEG, hoping that I could fool it by creating a new file, but it still didn’t work. I uploaded a photo of some pasta that I had posted on Instagram alongside the original — it caught the Instagram one. I tried downloading a stock photo, but it was also rejected. The only image it did accept were my original images taken on my iPhone.

I uploaded an original image of a kitchen cart, a screenshot of my dog and an Instagrammed photo of some pasta. Izitru only verified my original image.

I uploaded an original image of a kitchen cart, a screenshot of my dog and an Instagrammed photo of some pasta. Izitru only verified my original image.

While the software works well, the hard part will be motivating photographers to want to establish reliability for their photos on izitru — something I probably wouldn’t do for every photo I post of my dog.

“There’s a little hurdle there in that the photographer has to care enough to decide ‘I’m going to upload it here,'” Connor said. “It’s easy enough and it’s free, but still, there’s inertia and people are used to doing something a certain way.”

While the photo app may be great for one-off validation of breaking news photos or other images that are called into question, there is potential in the developer API, which isn’t free, for online dating sites, auction sites or citizen journalists. A dating site could automatically reject images if it detects that photo has been photoshopped. An auction site could offer a high degree of trust that the image photographed is the real deal. Websites like CNN’s iReport could check before they’re even published that the image from the the news event is an original, untouched photograph.

Still, if you want to check the validity of a too-good-to-be-true photo your friend is sharing on Facebook or Twitter, you can’t just download it and check because it will most likely fail izitru’s forensic testings. Instead, the image has to come from the original source for it to receive izitru’s trust rating.

“The bar we set is really high,” Connor said.

This article was updated on May 12 to correct the founder’s last name. It is Connor, not Conner.