This is the cocktail of disaster. On the one hand, a year containing an unprecedented number of elections in the world. United States, European Union, India, Brazil… more than half of the world’s population will be called to the polls in 2024. On the other, new generative AI which makes the production of false information more childish than ever. While social networks such as X are relaxing their moderation efforts, it is more necessary than ever to improve the reliability of the information circulating on these platforms. Director of operations of the Forum on Information and Democracy, Camille Grenier explains to L’Express how to get there.
L’Express: The framework that should be placed on social networks is complex to determine, because it touches on an important point: freedom of expression. How to think about this delicate question? Should we provide rules different from those of the physical world in the digital space, due to its amplification mechanisms?
In current debates, we forget an important point of Article 19 of the Universal Declaration of Human Rights: it puts freedom of expression on an equal footing with freedom of opinion. We can think of the latter in several ways. In 2024, it would be wise to think of it as also being the freedom not to be manipulated. The debates focused a lot on the freedom of the originator of an idea. But in the age of social networks, it is also important to ask the question of the freedom of the recipient. Because platforms can see what state of mind we are in and, based on that, push us certain types of content rather than others.
Will 2024 be a high-risk year in terms of disinformation?
Indeed, there will be elections in the United States, the European Union, India, South Africa… Half of the world’s population will be called to the polls. What’s more, with every crisis or moment of tension, the issue of disinformation always comes to the fore. It is more important than ever to work to protect the integrity of information. And to think about the obligations that we give to the platforms for this.
Is the architecture of social networks centered on the virality of content fundamentally problematic?
It can have positive effects, but it must be supervised. Particularly in crisis situations, for example conflicts. We can take inspiration from the circuit breakers that the financial markets have put in place. When a value gets too carried away and passes certain thresholds, we temporarily put it aside. On social networks, this could take the form of a pause, while human moderators study the post and put it back into circulation if they judge that it does not pose a problem.
Has the attitude of States towards social networks changed much?
In 2018, we were indeed very far, in Europe, from the spirit of the Digital Services Act (Editor’s note: European regulation which governs online platforms). We were still marked by the Arab Spring. There was this idea that freedom of expression could only be the bed of democracy. We are now less naive regarding the abuses that digital platforms can fuel. The regulation of social networks is no longer scary, but it must be framed by solid democratic institutions.
Can social networks, in practice, significantly improve the authenticity of the information circulating in their space?
Indeed, many actions can significantly improve the quality of content. The first step is transparency on the concrete means allocated to moderation: knowing how many humans in each country are tasked by the platforms with checking content that raises questions and, if necessary, deleting it. It is essential that there are systematically people who master the language of the country and know its cultural context. Not everything can be based on automated moderation, especially in conflict contexts where the decisions to be made are complex. In countries considered secondary markets by the platforms, there is unfortunately often no human moderator.
The other big challenge for improving social networks is to have information integrity criteria. They are not asked to distinguish truth from falsehood, but rather to distinguish sources of information that implement the ethical and professional standards of journalism. Basic criteria, such as verification of information from sources, the existence of correction mechanisms, transparency on the identity of the owners… Then promoting these sources, in their curation model. Reporters Without Borders (RSF) has launched the Journalism Trust Initiative (JTI), a standard that identifies sources of information that respect certain ethical and professional standards. This standard has been included in the EU’s strengthened code of practice against disinformation which is attached to the DSA. For the moment, there is unfortunately little action by platforms in this direction because it affects their curation model and therefore their economic model. Microsoft is the only company that has signed this clause.
The approach is interesting at a time when generative AI will make the production of fake news easier than ever. If they become too numerous for us to reasonably fight against, it is ultimately better to take the problem in the opposite direction and find how to revalue the verified information?
Indeed, if we compare the reach (Editor’s note: the number of people who will see the publication) of disinformation content and content that fact checks it, the second is generally much less than the first. We must therefore change the model. Reliable content today competes with any other content, even though it requires a much greater investment. We must therefore find suitable promotion mechanisms in order to give them back a competitive advantage.
A new generation of so-called “decentralized” social networks” are emerging at the moment, with the idea of no longer locking members into a space but of letting them interact with those of competing platforms (interoperability) or of changing easily, by taking their contacts and content (the data portability). Could they improve social networks?
We make, in fact, a lot of recommendations around interoperability and data portability. We can clearly see today that some of the users of X (formerly Twitter) are no longer satisfied with the experience but have difficulty changing services because they have a community of subscribers established there, and Rebuilding everything from scratch on a new platform is not easy. Will these “decentralized” networks improve or complicate the dissemination of quality information? However, it is a question that we are studying and which remains open for the moment.
.