Let’s address the issues of deep fakes

By Lars Seier Christensen,
Founder and Chairman of the Board, Concordium

I experienced the early years of the internet up close and, until a few years ago, considered this development to be the most significant innovation of my career. It was a transformative technology that some companies quickly adopted, while others failed to adjust their strategy in time, in some cases with fatal consequences.

However, I am convinced that we are currently in the midst of a much larger and more comprehensive disruption of both the business world and our private lives. Artificial intelligence, AI, is another transformative technology, but this time the development is happening at an exponential rate and with a significantly greater ramification. It is an inevitable technology that will have a decisive impact on how we conduct both our businesses and personal lives. AI will also have a significant influence on our private lives, including the way we convey and consume communication. New and revolutionary solutions to our problems, hopes, and dreams will be devised. In unison, new challenges will present themselves, as the recent debate in Denmark about a deepfake video of the Danish Prime Minister, Mette Frederiksen has shown.

Some of the first early adopters of new digital technologies are cybercriminals. They don’t care about rules and regulations, so they can operate their “business” undisturbed, as long as politicians and regulators do not convey solutions that can hold them accountable. And incidentally, provides consumers with transparent tools that enable them to distinguish between what is real and what is fraud.

Unfortunately, so-called ‘experts’ and politicians have managed to present arguments, criticisms, and proposals in connection with the debate on the deepfake video of Mette Frederiksen, which will make cybercriminals rejoice. Politicians can be partly excused, as it requires a deeper technological insight to rationalize effective and secure solutions, but some ‘experts’ ought to go back to the drawing board. Most proposals from that side seem to completely lack the basic professional realization that in the future one can only trust remote content that is properly verified from a reliable source.

Regulation and rules alone mount to absolutely nothing as long as offenders cannot be identified and eventually prosecuted. Simply declaring content as AI-generated — as has also been proposed — won’t amount to anything either, as virtually all content will be touched by some kind of AI. This is roughly equivalent to having to declare a website with: “This is a website”.

The most effective protection against deception and fraud is that offenders can be identified and that consumers can verify the origin of content — not whether the content is produced with the help of an AI tool. In the example of the deepfake video of the Danish Prime Minister, Mette Frederiksen, people themselves should be able to identify the video as a parody, if they could unequivocally verify that it was released by the Danish People’s Party. When the origin of content can be confirmed at any time, few thereafter will be able to claim ignorance of the video’s intention and authenticity.

The deep fake video of the Danish Prime Minister Mette Frederiksen published by the Danish People’s Party featuring the Danish MP and former MEP, Morten Messerschmidt https://youtu.be/WYIeGRV2pGU

Although the underlying technology is quite advanced, the most secure verification solution is quite simple and somewhat reminiscent of what we know from MitID, the Danish national eID. It could even be integrated into the same app, so we avoid having to install multiple apps. If a consumer wants to confirm the origin of a video, it can be solved just as easily as the way we use MitID.

The underlying principle — and I apologize for the technical jargon — is a so-called “Self-Sovereign Identity”, that is formalized using Verifiable Credentials and Decentralized Identifiers on an open and immutable blockchain. Such an ecosystem could serve as a foundation for how future consumers take control of their digital lives and their data, away from the prying eyes of Big-Tech and governments. The same technology has for years presented the very best option for a reliable, safe, and privacy-respecting age verification system politicians have so passionately craved for, but which yet again has eluded the ‘experts’.

Fortunately, we also have experts with the necessary technical insight to devise solutions that meet the requirements for accountability and security, while respecting fundamental privacy rights. As an example, the following was put forth in the Danish engineering journal, Ingeniøren:

“We need to understand that we can no longer trust images. And we can’t legislate our way out of that. But what we can do is have some systems that verify the sender and engrave the images,” says Sten Schaumburg-Møller, professor Dr Jur. at the University of Southern Denmark, who points to blockchain technologies as a possible tool for certifying that a video is authentic.

A great responsibility falls upon platforms, such as Facebook, X and YouTube, to facilitate technology that can identify content publishers, thereby verifying the content, so consumers can confirm its origin. This can be achieved, plainly put, by connecting the platforms’ video and audio players with the content publisher’s ID on a blockchain through advanced cryptography. In effect, the platforms must ensure that content publishers can be identified, while the responsibility for the content itself should lie with the content publishers. This also follows fundamental legal principles concerning accountability.

It is now up to the legislators to devise the necessary frameworks for a responsible use of AI, instead of outright banning the technology, thereby giving the cybercriminals free reign. Denmark is one of the world’s leading countries in key parts of the ultimate solution, namely cryptography and blockchain. So let us utilize our country’s potential and take the lead.

Notes:
Danish article in Weekendavisen here
The Danish debate about deep fakes is explained in English here
The full article from Ingeniøren here (Danish only)