It was already hard to have a civil exchange of ideas online. Comment sections and social feeds are full of anonymous strangers yelling at each other, and the political and cultural polarization in the U.S. has amplified the echo chamber effect. Add fraudulent photos and videos to the mix, and a fraught situation grows worse: how can opposing sides argue productively if they can’t even start from a shared understanding of what is objectively true?
This rise of deepfakes has kicked off a technological arms race between bad actors intent on spreading misinformation and those fighting to establish clarity. Sherif Hanna, VP of Ecosystem Development at Truepic, believes that instead of trying to get better at spotting sophisticated fakes, the better strategy is to establish which photos are true. His company has developed a blockchain-based solution to verify the provenance of images at the moment of their creation. On this episode of Mastering Innovation on SiriusXM Channel 132, Business Radio Powered by The Wharton School, Hanna explains why creating an open coalition of partners will be the only way to scale their solution across the internet.
An excerpt of the interview is transcribed below. Listen to more episodes here.
Saikat Chaudhuri: How does it work, technology-wise, that you’re able to verify [a photo or video] as authentic?
Sherif Hanna: Essentially, what we try to do is to get as close as possible to the very instant that the image is created by the camera. Once the image is produced — once you hit the shutter button on your phone — we generate the equivalent of a digital signature, a cryptographic signature, for that image. It’s a fingerprint of a sort which is unique to that photograph. We store a permanent record of this digital fingerprint. We use the blockchain for that, so there’s an immutable record of exactly what the image looked like as it was coming out of the camera.
Additionally, we verify the other data coming from the various sensors on the phone. So, for example, the photo is not just about the pixels of the image; it’s also about the date, time, and location, right? Where the image was captured, the context of the photo, often matters just as much as the pixel content of the photo. We go through over 20 tests of verification.
So we’ve verified the date and time when the photo was captured, we’ve verified that the location is not being spoofed by the phone, that this is the genuine location where the photo was captured, and we create a record of all of this information and the results of all these verification tests. If you receive a photo that was captured using Truepic’s technology, you can view a certificate of authenticity that what you’re looking at is exactly what came out of the camera, that the date and time has been verified, the location has been verified, etc.
We do that in two ways today. One is via a free app for both Android and IOS that anyone in the world can download and use. The other is by an end-to-end virtual inspection platform that we offer to our enterprise customers — insurance carriers, lenders, warranty companies — that require trust and documentation from their customers to conduct business.
Chaudhuri: To help our listeners understand, it’s more that the app needs to be present in the device in order to certify the authenticity and capture the attributes of the original, and not so much that I go afterwards and try to verify if an image is authentic or not. Correct?
Hanna: That’s exactly right. It’s about proving the provenance of a photo or video right from the instant that it’s captured.
“The future is not about detecting what’s fake, but rather proving what’s real.” — Sherif Hanna
The larger issue that the internet is facing is that if you receive a photo or a video and you don’t know where it comes from, there’s essentially no way for you to know whether it has been manipulated in some way. There’s been a lot of research into detection mechanisms that would look at the pixels of the photo and try to guess whether it has been edited or not. But given the rise of AI algorithms, the possibility of detecting manipulation is actually shrinking over time, given that AI is getting so good at forgery as well. There are also other manipulations of date, time, and location of the photo which leave no trace whatsoever. And so, the approach in the future, in our opinion, is that the future is not about detecting what’s fake, but rather proving what’s real.
Chaudhuri: I like that conceptualization. It makes a lot of sense to me.
I think one of the prerequisites and challenges for you is how to become the de facto standard in the industry. So everybody uses the technology and it’s just there, much like we have in terms of payment verification or something like that, something just built in right from the beginning.
Hanna: Yes, that is our long-term plan. We have actually started a partnership with Qualcomm Technologies who makes, basically, the brains inside a good portion of the smartphones that ship around the world. We’re getting our technology integrated such that it can run natively on the Qualcomm chip, so that any smartphone maker that uses these Qualcomm chips can then embed this as a native capability in the phone. When you as a user of one of those phones fire up the regular camera app that’s built into your phone, you can capture these verified photos and videos right there. You don’t have to download an extra app or anything like that.
Longer-term, we want to expand this and build an open consortium of companies that are involved in the capture and serving of photos and videos — and that’s essentially the entire internet — that would come together and agree on open standards for how to do this. We think that if this technology remains completely proprietary, it won’t scale, that really the only way to scale something to the size of the internet has to be driven through open standardization. That’s our plan for how we will proliferate this beyond just our implementation.
“There’s recognition broadly across the ecosystem that this is an imminent threat to the health of discourse on the internet.” — Sherif Hanna
Chaudhuri: Really powerful. I want to understand the role of partnerships in doing that, because clearly that’s not something you can do on your own. You have a partnership with Qualcomm. Who else are you engaging in partnerships with, or are you planning to have partnerships with, to really move this forward and open this up, but also set the standard?
Hanna: If you were to think about from the time that light hits a camera sensor to the time that that photo is displayed in an app or a browser, the image traverses a wide array of products, companies, and technologies. Our plan is to build a coalition that includes image sensor makers, the makers of the chipsets that are inside smartphones and other image capturing devices, operating systems vendors, online services, web browsers, and app developers.
It really has to be an all-up effort across the entire ecosystem in order to really get this going and get it to scale. And we don’t want to be the only ones doing it; we believe that there’s recognition broadly across the ecosystem that this is an imminent threat to the health of discourse on the internet. Therefore, there’s strong motivation from all parties to figure out a solution.
About Our Guest
Sherif Hanna leads Truepic’s efforts to proliferate verified photo and video technology to every image capturing device and app, to help restore trust in visual media. Prior to Truepic, he held various engineering, product management, and marketing leadership positions at Qualcomm, Atmel (now Microchip), and Cypress Semiconductor.