The online world is currently plagued by Deepfakes, malware, as well as fake news and more. It is now more important than ever to find ways to establish identity and trust. The data economy must seek ways to build on data and trust.
Digital tools are becoming more sophisticated as the world of technology evolves. Today, altering photographs, as well as videos, can be done with so much digital sophistication. The pinpoint accuracy with which this is done makes it difficult to sometimes identify whether they are authentic or not. Almost five billion people are predicted to become mobile phone owners. The proliferation of gadgets will also see the widespread use of laptops, tablets, security cameras, and more. All of these devices will surely be equipped with sensors that enable users to capture and transmit images and videos with lightning speed to the internet. It is very easy to share digital files randomly among millions of users.
The rise of Deepfakes
We easily trust whatever we see in videos as well as what we hear in audio recordings. Artificial intelligence now makes it possible to recreate faces of voices with a high level of accuracy. What we get eventually is a deepfake. Deepfakes are basically impersonations which are mostly used to misinform, create memes, or act pornography.
When you consider a deepfake of Nicolas Cage as well as the deepfake PSA of Peele, it is easy to see how strange this kind of technology is. Although they seem harmless, they make us wonder what the future would look like. Will it still be possible to take videos and audios for what they are? Can they still be trusted?
Deepfakes are new and making them is quite easy. Just a couple of years old, it has exploded and captured the interest of everyone in an intriguing way. Using artificial intelligence to recreate the looks or voices of people is actually quite captivating. And even if you have an old computer, all you need is the right software to create Deepfakes.
But they are doing harm to the real world
Deepfakes that are mostly available now have to do with memes, celebrity porn, as well as public service announcements. For now, they look quite harmless and you can easily identify them. But there are times when Deepfakes have helped to disseminate wrong information and cause damage in people’s lives.
With technological advancements, Deepfakes will continue to be refined to become more convincing. Think about the ill purposes that this technology will serve. With all of these in mind, what measures can be taken to ensure that what we share and consume online is trusted?
Intertrust and Deepfakes
Intertrust’s research and development project may well be the answer. At an event centered around blockchain and held in silicon valley, David Maher who is the current CTO of intertrust discussed the company’s project with the codename TIDALs.
David Maher, Ph.D., is Executive Vice President and CTO at Intertrust. He is considered a global expert on secure computing and is responsible for research and development at Intertrust, and also serves as Co-chairman of the Marlin Trust Management Organization which oversees the world’s only independent digital rights management ecosystem. Maher holds dozens of patents in secure computing; has published papers in the fields of mathematics and computer science; and has consulted with the National Science Foundation, National Security Agency, National Institute of Standards and Technology, and the Congressional Office of Technology Assessment.
David Maher further explains this project in this Q&A :
Q: What role should be played by centralized certificate management authorities in meeting the challenge of deepfakes? How do you envision the transition away from current infrastructure toward TIDALs?
A: “Current certificate authorities were designed to certify the identity and roles of a limited number of entities. When a few authorities are needed, a few CAs can fill the need. TIDALs make it possible to record authoritative knowledge from many different kinds of authorities regarding many different kinds of topics, and certify the provenance of many different kinds of events. TIDALs can be very flexible about topics and what can be said about them.”
Q: You describe TIDALs as “assertion-oriented blockchains.” Can you elaborate on this concept, and how does it differ from other applications of enterprise blockchain technology for managing secure identity?
A: “Other types of blockchains are used to record transactions, and assure that the transactions are valid (no double spending, inputs and outputs balanced, etc). Assertion-oriented blockchains merely record assertions made by authorities on specific topics. Here the terms authority and topic are quite abstract: an authority is an entity that is a first-hand witness of an event, or has definitive knowledge about an attribute of another entity. A topic is a category of knowledge that can include identity, but can include binding events with participants — like this photo was taken by this camera, by this photojournalist, at this time and place.”
Q: What organizations would act as authorities to build and maintain the TIDALs blockchain?
A: “First, the TIDAL system consists of many different ledgers each specialized to areas of authority. It is not one big chain, and therefore each TIDAL can have different policies for verifying submissions for different kinds of assertions, and different policies for agreeing on what entries go into the next block. This allows a device manufacturer to maintain their own TIDALs about the properties and identities of their devices, or they can submit their information to an entity operating a TIDAL supporting their kinds of devices. News organizations can do likewise, regarding assertions about provenance and chain of handling of video clips. The idea is that different organizations will be motivated to maintain truth and identity about their members or their products. There will be a natural organization of first-hand knowledge. Licensing agencies, regulatory agencies, guilds, trade groups, what have you, should find it easy to maintain or sponsor TIDALs that are relevant to their industries and operations.”
Q: You have described yourself as a skeptic when it comes to blockchain. What led you to blockchain as an answer to “deepfakes”
A: “Our research group saw that assertion-oriented blockchains could be configured in a very efficient way to immutably record real events and the authentic output of billions of sensors, binding them to entities that are responsible for them. We had designed a simple hash-based indexing system that makes verifying assertions to be extremely quick — quick enough to support verification of millions of video streams an hour. This is all possible because assertion-oriented blockchains have much less overhead than transaction-oriented blockchains, and hash indexes are much faster than certificates for verification.”
Q: What do you see as other emerging applications of TIDALs?
A: “The Internet, with IPv6, is increasingly used for all kinds of process automation for all kinds of industries, and literally millions of decisions will be made each hour and eventually each second based on information that is delivered to Internet-distributed applications in the IoT. Those decisions will be made with input validated using something like the TIDAL system of assertion ledgers. The security and safety of our infrastructure will depend on a highly efficient way of validating all sorts of entities and events. Decisions about the validity of photos and videos is just one example. Others will require greater scale, less latency, and increasingly better security.”
What TIDALs does is to fuse blockchain together with trusted assertions. This helps to come up with the technical foundations needed for a digital trust framework which actually help to tackle the problem of trust and other challenges.
Trusted Immutable Distributed Assertion Ledgers is designed to create a distributed framework which records and queries trusted assertions with respect to data and devices. This way, it becomes easier to decide the trustworthiness of these devices and sets of data. Systems that are based on this infrastructure will easily be able to determine a deepfake from an original piece. Finally, Intertrust might be the answer to identifying and defeating the rise of Deepfakes.