Swear’s founder came to Ventive to solve a problem he was passionate about—deepfakes and their impact on the social and political landscape. Turning his idea into a working product gave him the proof of concept he needed to venture into commercial implementation with a similar pattern.
In 2021, the FBI recognized deepfakes as an emerging danger that could undermine public trust if used by foreign adversaries or malicious actors. Deepfakes are videos or images of people that are digitally altered. This can include altering both facial features and audio. Deepfakes use both machine learning (ML) and artificial intelligence (AI) to manipulate audio and visual media believably.
For years, Hollywood has used versions of deepfakes to resurrect dead actors and superimpose their faces on live ones. With technological advancements and easy access to apps enabling anyone to create a fake video, law enforcement is increasingly concerned about the existential threat posed by manipulated media.
In the last five years, the quality of deepfakes has improved considerably. Some of the past clues that a video was fake have been removed by new technology. The unblinking stare and unnaturally still body has been replaced by microheartbeats under the skin that mimic humans uncannily. Currently, altered videos and media are so sophisticated that they’re undetectable by humans. In the next few years, there is a very real possibility that they’ll be undetectable by machines as well.
Forensic defense methods were first used to unearth deepfakes. But Swear’s founder knew that even though the average citizen had access to apps that could create deepfakes, they didn’t have access to detection technology. Additionally, deepfake technology was changing so fast that defensive techniques couldn’t keep up. When they were first created, deepfakes needed 30 minutes of pre-recorded videos to produce a semi-believable video. Today, an incredibly realistic manipulated video can be created after only seven seconds of a pre-recorded video.
Big tech’s approach to solving the problem of deepfakes involves creating complex detection utilities and programs. Instead of taking a defensive stance by constantly chasing after new technology, Swear wanted to be on the offense. They came to Ventive to build a product that would substantiate a video’s authenticity as soon as it was recorded. And they wanted an app that showed if a video had been altered and how it was changed.
The Swear app creates cryptographic fingerprints on a distributed blockchain ledger that authenticate video frame-by-frame as it's recorded. Attribution data is measured, which includes network, time, location, device, and sound bit. The attribution data verifies that the video was recorded by that particular device when and where it says it was, and that the audio hasn’t been changed.
Using Web3 technology, the app protects the truth and creates unbreakable digital content. Users download the app and record videos directly through the app. The video is protected as it’s created. The video is hashed on the blockchain as it’s being recorded. That makes each frame unique and verifiable.
A user can then choose one of three options. They can keep the video locally on their phone, choose to share it with select people, or they can upload it to secure cloud storage. Videos in the gallery can be listed as private or public, or they can be unlisted.
At the bottom of the video, viewers can see at a glance if the video has been altered in any way. If the bar is green, the video is untampered. If the video contains missing or unavailable data, the bar will be yellow in that section. A red bar shows which portion of the video is altered.
Six patents later, the Swear app is ready to launch for iOS and Web. The founder has proof that his idea is not only possible but can solve the root of the deepfake problem. Manipulated videos don’t need to be detected. Instead, by recording through the Swear app, video creators can prove that their videos are legitimate without relying on difficult-to-find detection technology.
There are many different potential markets for this product—in both the private and public sectors. In the private sector, content creators, citizen journalists, and activists can use the app to increase public trust in their content. In the public sector, security companies, military branches, or election officials could use the app to cut through misinformation by presenting verified media to the general public.
With the groundwork laid to scale up as much as needed, Swear is prepared to launch as the first tool of its kind to combat deepfakes. Ventive is proud to bring this product to market.