BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Will The Kinescope Rise Again In The Era Of Deep Fakes?

Following
This article is more than 4 years old.

Getty

Smartphone-based digital signatures have emerged as one of the most popular quick-fix solutions to combat the coming era of so-called “deep fakes.” If every image and video clip captured by a smartphone was digitally signed using a cryptographic signature unique to that individual and GPS-tagged with the precise location it was taken, the era of deep fakes would be over before it began, so the thinking goes. While such signatures present a wealth of privacy and safety concerns in the hands of repressive governments, their proponents have also failed to understand how trivially they can be overcome. In an era of ironclad digital media signatures, the kinescope is likely to rise again.

Once upon a time, in the early days of television, most broadcasts were performed live due a lack of recording technology that would allow them to be preserved and replayed later. Over time the kinescope emerged as one of the most popular recording technologies of the pre-videotape era. At their core, kinescopes represented a simple and obvious idea: point a film movie camera at a television screen and preserve it to film, whereby it could be played back later by pointing it into a television camera. In reality, the machines were far more complex, integrating technology to work around raster scan lines, different frame rates and the like, but at their core, kinescopes represented the simple idea of recording live television by literally filming a television set.

In fact, decades later in the early days of computer graphics, similar machines were used to record those images to still and moving film by pointing a traditional film or movie camera at a high resolution computer monitor inside a dark box using special lenses.

Fast forward today and we have cheap and even free do-it-yourself virtual reality headsets made out of cardboard and using cheap plastic lenses designed specifically for smartphones. These headsets and their accompanying lenses are designed to hold a smartphone securely in place for a person to view, but it would be straightforward to design a similar device that can hold two smartphones together, with a plastic lens connecting the camera of one to the screen of the other in a kind of simplistic home-built cardboard kinescope.

Armed with this cardboard kinescope, a person could simply walk to wherever their deep fake video claims to have been shot, put their cellphone into the box with a second cellphone (perhaps a borrowed phone or a cheap used one) and play back the deep fake video on the latter while filming it on the former. The end result is a digitally signed and GPS-tagged video fully verified and validated as being filmed at the claimed location using the claimed mobile device in real life.

The widespread availability of cheap and free cardboard VR headsets suggests this kind of “cardboard kinescope” could be trivially manufactured at scale and available all across the world. All it would take then is a few dollars and an account on one of the many microtasking Websites to pay people anywhere in the world to take one of these devices and film the requested deep fake video at the requested location. A person living in Chicago could craft a deep fake featuring the Eiffel Tower in the background, then use a microtasking site to commission a person living in Paris to grab their phone, a $1 cardboard kinescope and a borrowed second phone to play the deep fake, walk to the Eiffel Tower so the GPS coordinates are correct, play back the deep fake while recording onto their own phone, then post the resulting video, fully signed and verified as a “real” video.

With a bit of creative lens construction and a few automated playback filters to add the necessary hand shaking and visual imperfections, such a solution could actually work quite well.

Of course, phone manufacturers could defeat this process through new imaging techniques that incorporate additional sensors to capture the 3D contours of a scene, but likely as those techniques emerge new approaches to defeat them will emerge as well, perhaps lenticular constructions that reach back to the early era of 3D images or to early high-speed color film techniques.

Putting this all together, for every “quick fix” solution to problems like “deep fakes” there will always be a “quick defeat” solution that works around it. At first glance, the idea of person-specific digital signatures and GPS stamping sounds like a perfect instant solution to stop deep fakes in their tracks. Unfortunately, a quick combination of a cardboard kinescope and a microtasking site could trivially defeat such measures.

In the end, perhaps the most important lesson is that the only true way of combating deep fakes is to reestablish provenance to the digital marketplace of ideas combined with information literacy to begin the process of inoculating the public against the most egregious digital falsehoods.