How Digital Provenance Can Combat Disinformation
Disinformation has become an increasingly complex problem in our contemporary (social) media environment. The advent of deepfakes and other synthetic video tools makes it hard to believe what you are seeing online. This can impact industries from journalism and politics to insurance.
Digital Provenance is our strongest tool in the arsenal of disinformation prevention. By being able to verify the origin of a video or photo we can tell if it is real or synthetic. By leveraging Digital Provenance, we are one step closer to living in a world with less risk of the impacts of disinformation.
Let’s dive into how.
The last few years have seen an immense rise in disinformation across numerous media landscapes. This is especially troubling when we look to inform the public at large and ensure everyone is consuming well-sourced information.
The impact of increasingly sophisticated disinformation campaigns is not just on the consumer of content but also on institutions more broadly: Health disinformation can lead to amplified public health costs; the cost of operating a media business increases when one needs to incorporate fact-checkers and reactive disinformation counters to maintain brand trust; democratic integrity and public trust in institutions can be fractured when people don’t know what to believe or if what they are seeing is true.
It’s safe to say that disinformation is a problem that must be combatted on both an individual level, as well as an institutional level. This article explores how better provenance infrastructure and cryptography can help combat some of the most dangerous contemporary information security threats.
How disinformation is created
Disinformation is effective because it tricks the media consumer into believing what they see is what is happening. One way this is done is by manipulating or editing media: bad actors use deepfakes to create realistic fake videos to trick the audience into believing someone is saying or doing something they haven’t. This is done when a synthetic video is created that swaps the likeness of one person with another, in an attempt to make the viewer believe what you are seeing is in fact a video of a person that it is not. An example of this is Deep Tom Cruise, who uses deepfake technology to impersonate Tom Cruise and copy his likeness (although in this case it’s for comedic purpose).
We estimate that deepfakes will be indistinguishable from real content in 2-3 years. When anyone can make realistic videos, why should anyone trust photos or videos on the Internet at all
Another common form of disinformation/misinformation is posting a video without context (time or place). A bad actor may try to make a protest look more violent than it is, by posting a video from an unrelated riot. Without knowing the origin of the video, it becomes extremely hard to verify when and where the video might have been taken.
Our current methods of reducing the spread of these videos are through human fact-checkers and A. I tools that look to analyse the content and estimate the odds of them being corrupted or misused. However, these tools are not scalable, often unreliable and can only disprove the information after it has spread (potentially having already caused societal damage).
The core issue here is that there is no way currently to prove the origin of any piece of media online. We don’t know who created any particular video, if it's been edited or where it came from. We need proactive proof of the origin of media.
Enter Digital Provenance.
What is provenance?
Rather than trying to determine if the content is real, modified or synthetic, the way to solve this is to use Digital Provenance.
This simply means finding a mechanism to reliably answer the question of “Where did this come from and has it been modified since then?”. Cryptography provides us with the tools that we need to do this in a verifiable and robust way.
Now the question that the consumer of media will ask is not so much “does this image seem to be correct?” but “does this image have provenance and if it does, do I trust the provenance?”.
Think of this as the Twitter blue tick for authenticity.
But what does this mean in practise?
If you take an image on a trustworthy device, we know you are not a bot and take action to record this provenance, then why should your image not be trusted completely? This makes no claim as to the narrative being true, or if it has bias or not, it does not suggest the content creator is trustworthy, merely that this is what they saw, here and now. This does not check the facts, or censor the content, merely records an individual’s expression of it.
The power of this proof is that it allows us to know that a photo or video is real and that it was taken when it was claimed. When it circulates, we can always open its origin to convince ourselves of its authenticity.
How blockchain provenance works
Blockchain at its essence a peer-to-peer ledger technology that uses a network of nodes rather than a single entity to verify each transaction that occurs on it.
So, if I want to verify the provenance of a picture now:
Currently in digital photography (the non-blockchain enabled version):
The most secure way to check the provenance of a photo is to find the photographer (think of him as a node) and ask him about the picture. If the photo taker is a trusted journalist, we can be relatively confident they are telling the truth. However, we rely on trust in the person (a trust-based verification process). We can also check the meta-data of an image, and run A.I analysis to ensure a picture isn’t edited or doctored but this isn’t this easy to defeat and takes time and money.
So when you can’t completely trust the source, the system falls apart.
Blockchain enabled digital photography:
When a picture is taken through the OpenOrigins app the Digital Provenance is recorded immediately on the blockchain (via credentials secured at a hardware level). We can verify the media file (photo, metadata, image depth data) is unedited and hasn’t been doctored.
If it is ever corrupted or edited, we will be able to tell it is no longer the original file, as the provenance data will disappear.
Blockchains verify the image through a peer-to-peer network. The minute the OpenOrigins app captures the picture, numerous nodes (computers) in the network all corroborate the existence of the image as being the original. There is no more single point of failure (as data stored on the servers of a single company, is). Instead, the blockchain can prove it indisputably. This is a ‘trustless’ system.
Why is this so important?
Our world has a cocktail of innovations that, when combined, create a seriously complex environment for modern media.
Deepfakes and editing technology grow more sophisticated while user-generated content (UGC) becomes increasingly valuable for consumers who consume global, topical and timely news. These competing forces mean journalists have to rely more heavily on a network of sources (that they may not know personally), but need better means to prove that what they share is actually true. The pace of this media cycle is ripe for exploit by bad actors, so disinformation more accidentally seeps into trusted news networks causing reputation damage and worse, misinforming the reader. The ability for UGC verification to become automatic using the technological power of the blockchain is one of the solutions to this problem.
Digital Provenance is impactful for more than just journalists and media consumers, it is beneficial for industry too. Insurance providers can use Digital Provenance to ensure that the images they are sent in claims reports have not been doctored, reducing insurance fraud and cutting in-person costs. Development projects can more accurately assess remote sites without having to fly a team to the location. Politicians and CEOs can record themselves with Digital Provenance stamps so we can have complete confidence that we are actually listening to their words when we watch them.
If we know the origins of the videos and photographs, we view and we can verify that they haven’t been edited or corrupted, we know that what we are seeing is actually happening.
In this scenario — What You See, Is.
Even though disinformation is complex and hard to eliminate, the blockchain offers us a solution to the problem of effective provenance which can limit the spread of falsified videos by increasing trust in what we see.
Technology cannot be the complete solution to disinformation -- the problem of disinformation is as old as communication itself. However, recent advancements in technology have supercharged this problem. A platform where anyone can effectively review and trust video and photo content is an important building block towards being able to trust what we see online.
We hope the internet will become a little more verifiable and a bit more real again.
OpenOrigins is deeply passionate about countering deepfakes, disinformation and internet distrust. If you want to learn more or be involved in our mission, please reach out to us.