Since photography is the simple process of permanently capturing light waves, it is considered a perfect tool to capture reality.  It is, at its core, no different than our eyes. Even with its known limitations ( like less periphery, less color bandwidth or less dynamic range), it has and continues to be used to capture and document the world. From journalism to science, via casual travel pics or celebrity sightings, a photo undoubtedly proves and confirms a point. Well, it used to…

With technological progress comes creative destruction. While film created barriers on how much a photo could be altered, digital pulverized those barriers into oblivion: a camera or even light waves are no longer needed to create a perfectly credible image. And even those photos that are still taken by capturing light waves via a camera can be so dramatically altered that the result has little to do with the original snap. In turn,  trust in images is being destroyed. A recent statistic stated that only 15% of the readers believed that the cover images are accurate depictions of the model in reality.

Photography furnishes evidence. Something we hear about, but doubt, seems proven when we’re showing a photograph of it.” Susan Sontag  in On Photography.

This is an issue because, of all our senses,  vision is the primary source of trust (thirty to forty percent of our cerebral cortex). If we see, feel and hear conflicting signals, for our brains, vision always takes priority. If we can no longer trust what we see, if we do not trust images, we are functioning against our own nature.

It is furthermore an issue, because in this age of deep learning, where vast amounts of photos are used to train A.I.’s, what happens if most, or even some, are false images? The results could be terribly skewed and in the case of self-driving cars, trains or boats, deadly.

In the battle between integrity and alteration, the latter one has decades of advantages. Photoshop, the mother of all image alteration tool, was created in 1990. CGI dates back to the mid-1970’s. Today, pretty much all photos taken by our cell phones are altered in some way ( computational imaging) if only to make them look good. During the same period, nothing was created to protect, guarantee and secure the integrity of original images.  Camera and cell phone manufacturers have yet to offer proof of authenticity features to certify a file original integrity.  Which is puzzling since they are the highest risk of financial consequences from this erosion of confidence.

Most of the time, alterations are just for aesthetic purposes as above ©Tony Eckersley

The good news is that others have taken notice and solutions are starting to emerge. Not only to protect the veracity and credibility of editorial images but for those used for dating, insurance or scientific purposes.

Since they currently can only intervene post-capture, sometimes many levels apart, when a file has already been shared and republished, they have to rely on the existing information to validate an image. One company, Serelay, relies on metadata to verify if an image is authentic. Location, determined by GPS or cell tower signal, can confirm if the image was taken where it claims to be from.

Serelay uses multiple geo location metadata to confirm authenticity of a file.

Serelay uses multiple geolocation metadata to confirm the authenticity of a file.

Unfortunately,  metadata is rarely available on images posted online ( most social media delete all metadata to save space),  so the company also uses pixel alterations. But here again, multiple jpg compressions can return false positive and trick the results.

No Exif data, not a problem for Google. Using 126 million images gathered from the web ( all with location metadata), researchers trained a categorizer so it can recognize any outdoor location just from elements in the pictures. It beat humans on a head test. While Google plans to use the research in an upcoming iteration of Google photo ( to help users tag their images), it could certainly be a formidable tool in helping define the authenticity of an image. Once properly located, an image could be matched to its Google Street cousin, for example, to detect if there are any missing or added elements.

But nothing beats sealing the image at the moment of capture.TruePic is one company that offers to secure image and EXIF metadata in the blockchain forever. Thus any alteration can be immediately detected. While it works as an app for cell phones, their hope is that camera manufacturer will soon see its value and implement it as a built-in feature. Sealing the authenticity of an image would no longer be an opt-in option but a mandatory step.

Truepic is an image certification app.

Truepic is an image certification app.

The stakes are high: We now experience fake photos daily.  Mostly in the form of politically motivated memes, which have damaging long-lasting effects on the opinions of unsuspecting voters.  Insurance companies, who rely increasingly on visuals to process claims, lose $40 Billion a year on frauds.  Ebay and other “items for sales” apps repeatedly suffer from false advertising.  Dating apps used malignantly with fabricated or stolen profile pics. The list goes on. The current lack of ethics is profoundly eroding our trust in photography. It is up to technology to restore it.

Share Button

Comments are closed.

Post Navigation