This is unheard of in Europe. One member state, Romania, had to invalidate a presidential election after being the victim of a political influence operation on TikTok. A galaxy of influencers and fake accounts boosted the popularity of a pro-Russian and conspiratorial far-right candidate: Calin Georgescu. So much so that this virtual unknown went from 1% of voting intentions to 23% of votes in three weeks without having held the slightest meeting. Faced with the seriousness of the matter, the European Commission opened an investigation against the social network on December 17. But what can we expect? And what should Europeans do to protect their elections from foreign interference? Interview with David Colon, professor at Sciences Po Paris, specialist in the history of propaganda and author of Information Warfare published by Taillandier.
L’Express: What do we know about how TikTok was used during the Romanian elections?
David Colon: Algorithmic cognitive warfare, that is to say exploiting the properties of network algorithms in order to influence perceptions, manifested itself in two ways in the Romanian case. The first, fairly classic since 2016, is the distribution of unauthorized targeted political advertisements on a large scale, on Facebook. Ads that escaped Meta’s detection capabilities. The second is the exploitation of the TikTok effect, that is to say the extreme sensitivity of the recommendation algorithms of this platform which make it possible to influence the content presented to readers. Concretely, influencers who probably had no idea of the final objective of the campaign were asked to produce content. Volunteers were recruited to write thousands of fake comments under this content. Thousands of fake TikTok accounts have amplified this. The result was that candidate Calin Georgescu’s videos were made virtually unmissable for anyone watching content related to the Romanian elections. While this candidate did not make a poster, public meeting, or interview, he went from 1% of voting intentions to 23% of the vote.
Do we have any concrete clues about the identity of the sponsor of this influence campaign?
Romanian intelligence has declassified a series of documents which lead it to consider that an operation of this magnitude and this coordinated can only come from a state actor. Evidence suggests that funding for the campaign was done through a Ukrainian company but it may have been used without its knowledge. The element which points, in my opinion, most towards Russia, was revealed by an investigation by Nicolas Quénel on Intelligence Online. It highlights the role of a company domiciled in London, but of Russian origin. Context also plays a role. We know that the Kremlin interfered in the Moldovan election, the Georgian election, the European elections as well as the French legislative elections. The Kremlin has interfered in all European elections. It is therefore highly probable that he did and continues to do so in Romania.
The European Commission has opened an investigation against TikTok. What types of items will she look for and what are her chances of finding them?
It should be borne in mind that this investigation is being opened under the Digital Services Act. The Commission is therefore not seeking to identify the sponsors of the operation, but to find out whether or not TikTok has respected the obligations imposed on the platforms in terms of management of certain content, particularly political content. It is easy to fool platform algorithms. In the case of Romania, the sponsors deceived TikTok’s algorithms by having influencers make videos encouraging people to vote, without indicating the specific name of the candidate, but by accompanying these videos with comments favorable to Georgescu. What TikTok is criticized for is not having put in place tools to detect this type of stratagem.
Could platforms effectively prevent these phenomena?
Yes, this is in theory a fairly easy problem to solve. TikTok’s recommendation algorithms are set to be extremely sensitive to the virality of content. The objective of the platform is in fact to offer the most recent, attractive and addictive content possible. It is therefore sufficient in theory for the platform to reduce the sensitivity of its recommendation algorithms to limit the risk of them being manipulated, but this would of course go against its economic model and its interests.
Facebook, Instagram… Don’t other social networks have exactly the same flaws as TikTok?
The economic model of social networks based on advertising leads them to maximize virality and try to keep Internet users on their platform as long as possible. Which mechanically produces harmful effects on users and societies. What sets TikTok apart from other social media is its de facto subordination to the Chinese Communist Party, through its parent company, Bytedance. This poses two major questions for the future of our democracies. Do we want the information that the majority of the French population will access tomorrow to be determined by algorithms over which the CCP can exercise influence? And do we want the information accessed by the majority of the population tomorrow to be so influenceable by coordinated algorithm manipulation operations?
Since its acquisition by Elon Musk, X has been filtering less and less problematic content, particularly that conveying disinformation. Elon Musk was also the first boss of a major social network to actively campaign in favor of a candidate in the presidential election. Is there not, here too, a pronounced risk of interference in the electoral processes?
X is a major lever of interference in the political debate because it counts many journalists, politicians and experts among its users. It therefore has a great influence on the way in which the media agenda is determined, and more broadly on public debate. On the other hand, based on available knowledge, X, with its hundreds of millions of users, does not produce the same effect on voters as TikTok, which has more than 1.5 billion users worldwide.
What should Europe do in the face of these new risks? Is the United States right to consider banning TikTok?
To begin with, all social media should be banned for those under 16, as Australia has done. Then, the experience of the ban on RT in Europe and then on TikTok in New Caledonia should encourage us to be cautious, given the ease with which such a measure can be circumvented, and its negative impact on confidence. citizens towards those in power. What Europe should do, on the other hand, is to urgently create a virtuous European social network, respectful of current European regulations and not based on an advertising model. Such a social network is essential today to protect our public space from foreign interference boosted by artificial intelligence and algorithmic cognitive warfare. Such a network could rely on three sources of income: public funding, private funding from companies aware of their democratic responsibility – at the same time as the risks that disinformation represents for them – and, finally, the contribution of its users.
With the Digital Services Act, does Europe have the adequate framework and resources to detect the failings of social networks and force them to respect its rules?
Everything essentially relies on the goodwill of the platforms, which are only exposed to financial sanctions. As long as the amount of profit generated by the offense remains greater than that of the fine, platforms will be encouraged to violate European regulations.
The Digital Services Act also allows a platform to be temporarily banned, which can have a heavy economic impact on it.
Yes, but this is only in the case of repeated convictions. And for the moment we hardly see any concrete results from the first investigations opened in 2023.
Do you think that the EU will never dare to use this weapon?
You never know. But in the event that it were used, I doubt that it would produce the slightest effect. Because there is a total gap between the technological pace and the political pace. The famous phrase of Lawrence Lessig, this Harvard law professor, sums it up perfectly: “Code is law” [NDLR : le code informatique fait loi]. The code produces an immediate and universal effect. This is not the case with any law.
Ban a social network who does not respect our rules leaves you skeptical. Is it because the ban seems inherently a bad idea to you or do you think it’s a good idea in theory, unfeasible in practice?
I think the good idea is to equip ourselves with honest social networks. We’ve been trying to control social media for eight years. Everything that was undertaken has failed. It’s time to think about another way of doing things. There is a lot of talk about European sovereignty today. It cannot exist without digital sovereignty. It is easier to create a European social network than to try to train dragons.
.