June 28, 2022

By Nicolas Hénin, Researcher at EU DisinfoLab

French fact-checkers verified a total of 169 disinformation items linked to the presidential election. Their analysis allows us to better understand what were the main narratives of the electoral hoaxes, their formats and characteristics, as well as the actors that spread and amplified them.

We have gathered all the articles published by the six main French fact-checking organizations (members of the IFCN) debunking the hoaxes that disinformed about the last presidential election.

As we selected all the articles that tackled disinformation, we finally established a corpus of 169 debunks related to the presidential election that circulated in France from 1 January to 31 April 2022 (several hoaxes concerned the reliability of the ballot and the vote count, which is why we included the few days following the second round).

It is important to note, beyond this volume, that none of these hoaxes reached a magnitude that could have altered the integrity of the voting process or jeopardised its outcome. Furthermore, no significant evidence of foreign interference was found in either the production or propagation of disinformation related to this vote. The impact of the invasion of Ukraine was particularly challenging to predict, making everyone wonder whether the conflict would expose the election to additional disinformation. On the one hand, Emmanuel Macron’s stated support to Ukraine could have possibly made him a prime target for a Russian operation, as he was during his first election. On the other hand, Russia’s capacity to produce and disseminate disinformation was contained and to a large extent dismantled in the early days after the war broke out.

1. Review of disinformative narratives

Overall, many election-related mis- and disinformation narratives circulated. The most widespread was political polarisation, which is expected during an electoral campaign since the candidates seek to present their program in the best light and stigmatise their competitors. Somewhat more surprising was the place occupied by economic polarisation, reflecting the focus on issues related to inflation and purchasing power, topics that have been traditionally seized by the far-right.

As a result, other recurrent far-right topics – namely immigration and insecurity – respectively dropped to 8.3% and 1.8% of the hoaxes that were debunked.

Figure 1. Distribution of disinformative narratives

The most problematic narratives, the ones aimed at undermining public confidence in the electoral process and its outcome, accounted for 12.4% of the items. This rather high proportion probably reflects the fact that they were especially scrutinised by fact-checkers. In practice, these narratives did not spread very widely, and failed to break through the barrier of the traditional media.

To use Ben Nimmo’s breakout scale on the impact of influence operations (see the graph below), these narratives have hardly reached category 4, or did so only occasionally.

Among these hoaxes aimed at spreading electoral distrust, there was:

  • A claim that 22.500 Yellow Vests would have been deprived of the right to vote by a 2020 law;
  • The sudden disappearance of 2 million votes for Marine le Pen during the election TV program on France 2 which would prove an existing fraud;
  • The claim that Emmanuel Macron was the number one candidate in “100% of electronic voting machines”;
  • The photos of slightly torn ballot papers that would aim to invalidate votes for Marine le Pen;
  • A nonsensical rumor that the QR code printed on the new voter cards was allegedly for the purpose of tracking voters or rigging the results;
  • A myth about American company Dominion, accused by Donald Trump’s supporters of having played a crucial role in the “steal of the vote”, which would have been commissioned to count the votes.

While conspiracy narratives trying to prove electoral fraud have had little success in reaching a large audience, looking at the profiles of the people sharing them on social networks shows a rather disturbing picture. These people, though few in number, seem to be deeply anchored to some kind of parallel reality and are sharing a significant amount of false news, a particularly high proportion of which is of a harmful nature.

This audience, which is a strong disseminator of disinformation and is disconnected from traditional information, presents a significant social challenge in terms of radicalisation and reintegration into the political game.

2. The more you speak, the more likely to get fact-checked: the paradox of disinformation and presidential candidates

Candidates were the source of the debunked items in 54.4% of the cases. This figure should be taken with caution, as it results from the fact that fact-checkers devoted a large part of their efforts to verifying the statements of the candidates, whose every intervention in the media was scrutinised.

Figure 2. Distribution of disinformation transmitters (main source).

In addition, we classified as a candidate any person who indicated his or her intention to run in the election, whether or not he or she had obtained the necessary sponsorship to actually enter the race.

In detail, the candidates mostly mocked their opponent or presented true elements in an exaggerated or biased way in order to justify their programme. Instead, the most problematic elements, especially those that could be considered conspiratorial, came mainly from anonymous, marginal and often extremist actors.

3. Videos as the major electoral disinformation communication supports

As videos remain a favorite medium of political communication, it is not surprising to find it as the support for almost a quarter of the disinformation elements in our corpus (62.7%), followed by pictures (16%).

This over-representation of video is a consequence of the fact that the campaign was heavily televised, but also that all media have been converted to video: radio shows, for example, are often shared in video format on social networks.

Figure 3. Distribution of disinformation supports.

4. Electoral disinformation typologies: misleading information is dominant

According to the typology established by First Draft, the vast majority (84.6%) of the disinformation elements contained in our corpus can be considered as misleading content, i.e. “the use of information to frame an issue or individual”. Fabricated content, which includes content that is entirely false, accounts for only 4.7% of hoaxes (8 in total).

Only two specimens of imposter content were reported (a fake tweet from RTL and a forged letter from the Les Républicains party). Six pieces of manipulated content were debunked, including a fake poll announcing Marine le Pen’s victory, incorrect quotes attributed to political figures who worry that the election will be rigged, and a doctored Swiss radio and television poll announcing Emmanuel Macron’s defeat.

Figure 4. Distribution of the disinformation items according to the First Draft typology.

Finally, it is worth noting the important work done by the media to dismantle the most viral or harmful rumours about the election in an informative and unbiased way. Libération, for example, published a long article, on the Dominion rumour, soon after its appearance. As for Le Monde, it produced an article and a podcast on the various suspicions of electoral fraud that were circulating.

5. Disinformation export to neighboring countries

As disinformation is an increasingly transnational phenomenon and events like elections have resonance beyond the country in which they occur, EU DisinfoLab researchers Ana Romero (monitoring Spain), and Raquel Miguel (monitoring Germany) have identified some hoaxes that circulated in the countries they monitor about these elections. Previous experiences, such as the spread of disinformation narratives about Covid-19 across Europe, had already shown us how narratives originating in one country could spread abroad, and sometimes even have more virality in another country.

The French elections have brought to Spain old repertoires on electoral fraud, false party proposals, or impersonation setups. It is observed that far-right supporters of Marie Le Pen were the ones promoting the electoral fraud narrative, with Telegram being the main channel to spread false messages in Spanish. The war in Ukraine was instrumentalised to share misleading content against Macron while launching anti-refugee stances. False post-election proclamations have also proliferated, such as the one attributed to Macron against leftist policies.

In Germany, one debunked piece of disinformation claimed that Macron met Klaus Schwab (chairman of the World Economic Forum) only one day after the presidential election in France. Another was supportive of far-right Marine le Pen, asserting that her popularity has risen sharply over the past ten years.

Two narratives fueling electoral distrust by pushing the electoral fraud conspiracy have been imported from France where they first circulated. A video defending that “Le Pen ballot papers have already been delivered invalid” and a screenshot from a TV program in which Macron is attributed 14.21 million votes and Le Pen 14.32 million. “The screenshot was taken 30 minutes before the end of voting. She was winning. 30 minutes later she was losing by a margin of more than 10%.”

France 24 Observers also published an article after noticing a decontextualised video circulating on an Italian Telegram channel. It shows pieces of paper piled up in a truck and thrown into the street, before the police intervene. For some Internet users, these papers would be ballots for Marine Le Pen, abandoned so that they are not counted. In reality, the video shows the action of a student association to denounce the links between the Rassemblement National and Russia.

6. Impact assessment

Kantar released a study to assess the impact that disinformation may have had on the decision-making process of voters. The researchers set up two groups representative of the population. The first group was asked to share their voting intentions, while the second group was first exposed to false news before saying whom they would vote for. The result is rather counter-intuitive: when voters are bombarded with intoxicating information that makes them think that the country is being mismanaged and is rushing towards collapse, they tend to turn to a person who reassures them and whom they believe can manage the crisis.

This phenomenon is particularly visible for topics dear to the extreme right, such as vaccines or immigration. This study was covered in a FranceInfo Vrai ou Fake article and video.

Conclusions and thoughts

  • Of the 169 debunks around the elections analysed, we found that political polarisation is the most recurrent theme.
  • Candidates were the first transmitters of debunked items, most certainly because they were constantly monitored by fact-checkers. Yet most of the debunks about them were largely inaccurate or exaggerated.
  • Disinformation widely resorted to visual elements such as videos, and mainly appeared in the form of misleading content.
  • Disinformation narratives around the French elections were also exported to Spain, Germany, and Italy.
  • The most dangerous narratives were those trying to undermine public confidence in the vote, but they remained quite marginal and mostly excluded from traditional media.