Fool me once: Russian Influence Operation Doppelganger continues on X and Facebook – French chapter

Posted on:  2024-09-02

This report is one chapter of a broader initiative coordinated by the Counter Disinformation Network. The initiative aims at offering a comprehensive view into the objectives, techniques and key messages of a Russia-aligned online influence operation, focusing on its activity on X (formerly Twitter) in 6 languages during June 2024. 
More details about the influence operation, the other countries it targeted, and the methodology can be found in the full report.

Report co-author: Florian Roussel, independent researcher

Summary

The French corpus (see Methodology appendix) suggests three main objectives for the influence operation:

  • Shaping the domestic discourse and political landscape, by attacking the current government and supporting right and far-right political alternatives
  • Influencing the discourse around the war in Ukraine, by discrediting Ukraine as a whole or its government (including personal attacks on Volodymyr Zelenskyy) framing support for Ukraine as being against the French national interest, and spreading messages in favor of Russia.
  • Destabilizing Western alliances, through attacks on the EU, France’s European partners, the United States or NATO.

These three lines of attacks can interact within the same tweet. For instance, one tweet paints Macron and his government as NATO puppets who are causing damage to France by sanctioning Russian gas and choosing expensive American gas.

In order to reach its objectives, the influence operation attempts to make use of domestic news or context, often trying to frame the tweets in the context of larger French political debates. 

For instance, one of the angle’s most frequent messages is criticism of E. Macron (a mainstay of the political discourse in France, where sitting presidents rarely enjoy approval ratings above 35%). Echoing sovereignist tropes, the campaign also sometimes makes appeals to an idealized, yet unspecified “French People” or “France”.

As detailed below, more specific news events are also used in the messaging, such as a controversy around the singer Adele, the shortage of a drug to treat allergies, or the financial markets’ reaction to the calling of snap elections.

* * *

Overall network activity

362 French-language posts (see Methodology appendix for details about their collection) were identified during the month of June 2024, with peaks recorded during the periods June 11-15 and June 19-22.

These peaks do not appear to evidently match particularly news-rich periods in France within our studied timeframe, suggesting that the peak activity of the presumed network orchestrating the campaign is mainly influenced by issues not necessarily pertaining to France alone. A weekly drop in activity on Mondays and Sundays also suggests internal organizational constraints influencing the activity, which does not match any known newsworthiness pattern.

Figure 1 – Daily number of tweets in French

Content and narratives

Overview

Overall, the narratives promoted by the tweets in the French corpus range widely but are in line with the objectives and modus operandi of previous campaigns attributed to Russian influence operations: stoking division inside the target country, sowing distrust between the target country and its international allies, undermining support for Ukraine (incl. through personal attacks against Volodimir Volodymyr Zelenskyy), or by promoting political parties perceived as friendlier towards Russia.

While these broad objectives appear consistent across countries targeted by Russian disinformation operations, the campaign makes clear attempts at “localizing” its content by referring to the French context, presumably so as to increase the likelihood that it resonates with genuine audiences.

Salient narratives

(The same tweet can promote more than one narrative.)

Attacks on Emmanuel Macron and his government

52% (188 out of 362) of tweets targeted the current French government. More than half of these posts (56%) criticized the French government on domestic issues (or in very generic terms such as “Macron has lost the confidence of the French people”), while about a quarter (28%) specifically targeted French support for Ukraine.

In addition, 9 tweets took aim at broader categories with anti-elite speech without designating a clear individual or political side, for instance targeting ‘politicians’ in general or lamenting the rise of ‘millionaires’ as a sign of an unhealthy economic system.

Messages openly expressing support for another party to take power in France accounted for 12% of posts. The only party specifically mentioned as a preferable alternative was the far-right Rassemblement National (mentioned specifically in 34 posts), which has long been accused of cultivating close ties with the Kremlin. A few other posts were less specific, identifying “the right” as the preferred political alternative.

Undermining support for Ukraine chiefly through indirect attacks on actors supporting it

In line with the Kremlin’s foreign policy priorities, the campaign’s likely overarching objective is to undermine Western support for Ukraine. However, the favored way to approach the objective is not to address it heads-on: over half (57%) of posts do not refer either implicitly or explicitly to any Ukraine-related topics and focus chiefly on polarized domestic issues and attacks on the current government.

Of course, given the pro-Ukraine stance of the current French government, any campaign aimed at weakening it is an indirect tool to undermine support for Ukraine.

Even when directly addressing Ukraine, the dominant narrative is focused on the costs to France

Out of the 156 posts that directly address the war in Ukraine, 79 attempt to frame an explicit contradiction between supporting Ukraine and the supposed French national interest. For instance, the campaign highlights the fact that the resources sent to Ukraine are sorely needed at home (17 posts), that the French people are being dragged into a war in which they have nothing to gain (14 posts) or that economic sanctions against Russia are backfiring (10 posts).

However, some posts directly address the situation in Ukraine without highlighting a French-specific angle, for instance : discrediting Volodymyr Zelenskyy (as incompetent or a warmonger), highlighting the issue of corruption in Ukraine (including Volodymyr Zelenskyy), framing the Peace Summit held in Switzerland as a diplomatic failure or promoting pacifist messages.

Russia portrayed as a responsible international power with a thriving economy

13 tweets focus on painting a positive image of Russia. Two narratives dominate:

  • presenting Russia as an economically thriving country (4 tweets), in spite of Western sanctions, from which France could draw inspiration,
  • Portraying Russia as a reasonable geopolitical player interested in peace and stability (6 tweets), likely as a reference to then 2-week-old overtures by Russia to freeze the conflict on the basis of the territories held at that point in time. 
So-called allies and supranational organizations are working against France’s interests

23 tweets (6.4% of the corpus) express in one form or another the sentiment that France is a vassal to the US, the EU and/or NATO. One noteworthy variant is that Macron and his government are willingly selling out the French national interest to those actors for unspecified reasons.

The dog that did not bark

Inexplicably, some other topics barely register in the dataset, although they were hotly debated in the domestic information environment during the month of June and could have offered greater opportunity to drive the campaign’s objective to drive polarization and undermine support for the French government. Among those topics:

  • Emmanuel Macron’s decision to call for snap parliamentary elections following his party’s defeat in the European elections on June 9 was widely described as irresponsible, including among some of his political allies (the corpus does not display any mention of the debate around whether dissolving Parliament was a responsible decision, although politics are widely mentioned),
  • Migration (both legal and illegal) was a central topic of the June election campaign, although it barely appears in the dataset (1 mention in the corpus),
  • Often linked to the above, the place of Islam in French society is an evergreen subject of polarization (no mention in the corpus),
  • Law and order, and in particular a number of tragic violent incidents that became points of contention in the national debate (in the corpus, 8 tweets mention public security as a problem in very vague terms, and a further 3 specifically aim to cast doubt over the security of the Olympic Games, without going into any specifics).

The relative underrepresentation of these topics is all the more surprising since some of them have been leveraged to a great extent by previous Russia-aligned influence operations (e.g. 1, 2, 3). One possibility is that the operation was not organized or informed enough to change its focus in line with shifting public debate, likely reducing its efficacy.

Qualitative insights into observed techniques and behaviors

Reusing the same images on different tweets promoting the same narrative

Some images surfaced multiple times in batches of posts that were variations around the same theme. For instance, an image from 2023 showing riot police in front of a restaurant where Emmanuel Macron’s campaign celebrated his victory in the 2017 presidential elections appeared in at least three tweets (see below) promoting the message that Macron enjoyed little support from the French population.

Figure 2 – Example of different posts promoting the same narrative using the same image. Left: “It’s incredible what Macron has done to our economy and social system”; Middle: “The right seems stronger and stronger. This is a sign of growing discontent towards Macron”; Right: “With Macron’s approval rates, I wouldn’t be surprised if the right won the National Assembly”.

Low-quality content

Despite technological advances in content-generation tools available to malign actors, a number of posts are clearly produced by actors lacking basic contextual clues. For example, a tweet which appears to be advocating for peace talks involving all actors displays an image mixing young people partying, Western and Ukrainian political leaders, and a large conference room table (see below).

Figure 3 – Example of a post whose attached image seems only largely irrelevant to the post’s text and to the broader narrative it supports. Translation:”Only a discussion between all concerned parties can bring peace”

Likewise, 6 tweets support “l’Union Nationale” as a political party. Such a party does not exist in France, but its semantic similarity to ‘Rassemblement National’ as well as the mention of Jordan Bardella, Rassemblement National’s leader, in a tweet promoting l’Union Nationale suggests that some automated translation or rephrasing technology was used by non-native French speakers.

Some (largely failed) attempt at surfing on the news cycle

Although some topics with important polarization potential appeared to be left out by the network (see above), some of the content posted made explicit or implicit reference to current events.

Announcements regarding the sending of French military instructors to Ukraine

For instance, on June 15, 5 tweets shared the same image of soldiers evacuating a body bag from a collapsed building in Kharkiv taken from the “readers’ comments” section of French periodical L’Express, as well as a link to the page. All tweets criticized Macron, presumably for his stated intention to send military instructors to Ukraine (despite discussions to that effect, as of August 29, France has not officially sent military instructors to the country, although they are reported to have been present since at least 2023).

However, the campaign appears to refer to statements made at the end of May and the beginning of June (June 6-June 9), approximately one week before the tweets were posted. This week-long lag between the newsworthiness of the information and the publications by the network points to slow adaptation by the campaign.

The singer Adele

Likewise, on June 20 at 6:34 pm, 4 different tweets expressed support for the singer Adele, who had been at the center of an online controversy for wearing a dress made by a Russian designer with ties to the Russian army during a performance on June 15. The tweets expressed support for Adele’s choice of dress (it is unclear if the singer was aware of the dress’ provenance). As above, given how late those tweets came in the news cycle, their impact on real audiences is likely to have been muted.

A financial “crisis” ?

Following the announcement of snap elections on June 9th, the interest rate on French government bonds increased, reflecting a rise in investors’ risk perception.  While significant, this increase was not cataclysmic, sending 10 year German-French government bond spreads back to mid-2022 levels. However, three tweets, all posted on June 15 at 4:10 pm, mention a financial crisis, overplaying the gravity of the situation in a likely attempt at playing on fears of economic decline.

Figure 4 – Examples of posts promoting the unsupported idea that France suffered a financial crisis. Left:”This financial crisis is an absolute disaster for France”. Middle: “I feel sick when I see the interest rates on bonds falling to the bottom”; Right:”Never seen such a drop in French stock markets. It is shameful”.

Drug shortage

French media reported in June about shortages of certain medicines including Ventolin, a widely used drug against asthma. The campaign seized this issue with at least three tweets posted on June 13, directly blaming the government for the shortages, playing into a wider societal debate over the state of France’s healthcare system. The posts misspelled Ventolin by missing an “e” at the end that is required in French. 

Figure 5 – Examples of posts, one of which pins the blame squarely on the government, lamenting the shortage of Ventolin; Left: “I’m allergic and I really need Ventolin. What’s going on in the government? Don’t they think of us?”; Right: ”How to even live during the floral period without Ventolin? It’s unimaginable for me!”

Ukraine Peace Summit

Three sets of tweets refer to the June 15-16 Conference on Peace in Ukraine, held in Switzerland. The first set was sent on the June 15, criticizing the absence of Russia or the supposed contradiction between having a peace conference while the war is ongoing. The second and third sets, sent respectively on the June 19 and 21, attack the conference’s alleged lack of results and Volodymyr Zelenskyy for allegedly being a warmonger.

A few pieces of content in the dataset offered no evident connection to known Russian interests. For example, a few posts express worry about plastic pollution (and all share the same edited image of a tap floating in mid-air and spilling plastic trash), which, to the best of our knowledge hasn’t been a recurring theme in previous influence operations linked to Russia.

Figure 6 – Examples of posts with unclear links to Russian interests yet presumably posted by the same campaign; Left: “If we don’t pay attention now, future generations will pay the price”; Middle: “Plastic pollution is changing our evolution. It’s terrifying!”; Right: “It’s alarming! We should take these ecological questions a lot more seriously.”

Attacking other Western governments

3 tweets in the dataset attack the authorities of other Western governments on purely domestic matters, without bringing France into the discussion: two tweets lament Germany ‘losing its human rights’ (possibly referencing the ongoing investigations into the far-right AfD party), and one attacks Hunter Biden as a symbol of corruption in the United States. 

Impact

The French corpus for June has recorded 1,152,364 views (on average, 3,157 views per post, according to official Twitter viewership data. However, the view counts of the posts in the campaign appear abnormal (batches of tweets posted around the same time all obtaining very close numbers of views) and are likely indicative of artificial views collections.

Replies and retweets, which do not display any evident sign of being artificially inflated, have much lower counts. In total, the 362 tweets generated 82 replies and 24 retweets, suggesting overall low campaign efficacy.

Recommendations

In line with Very Large Online Platforms’ obligations under the Digital Services Act, and in particular those stemming from the election guidelines (June 2024 being an election campaign month in France), we recommend that X (formerly Twitter) increase its investments in systems and partnerships to respond more effectively to foreign influence campaigns.  

In particular, given this campaign’s easy-to-spot inauthentic patterns (e.g. 1-2 thousand retweets and no comments per tweet, number of views that are too round to be random, accounts created to post just 2 or 3 tweets about political affairs then go dormant…), X should be able to put automated alerting systems in place for its moderation team to review.

X should also consider rejoining the Code of Practice on Disinformation to strengthen its ties with civil society and benefit from other stakeholders’ insights and experience into tackling disinformation campaigns.


Appendix – Methodology

The corpus analyzed in this report refers to the French-language subset of data collected as part of the overall effort cross-language effort at monitoring the influence operation’s language on X.

The full corpus was built to be keyword-agnostic and relied on running a search focused on structural properties of the posts: using a social listening tool, the query collected all posts on X bearing the paradoxical combination of >300 shares but <100 likes, coming from accounts with <100 followers, and containing a link and image or video.

The search period ran from June 4th to June 28th 2024 and yielded 1.366 posts, of which 362 were in French.

About the Counter Disinformation Network

The Counter Disinformation Network (CDN) is a collaborative platform that gathers more than 130 disinformation-countering practitioners from 40 civil society organizations, universities, news organizations, fact-checking organizations and independents mostly from Europe and North America. The network was initially convened by Alliance4Europe with the aim of supporting a whole of society approach to countering disinformation by bringing together the different parts of our community of defenders and finding ways to work together. All members of the network can propose projects, using the network as a force amplifier to enable research that would not be possible individually. We believe in not reinventing the wheel. Therefore, CDN aims to feed and support existing initiatives, such as the FIMI ISAC and the DAD-CDM project. The network seeks to help mobilize resources with different experts and from different countries to write comprehensive reports, highlighting systemic issues and foreign information manipulation and interference, and working together to promote it across a wide range of countries and stakeholders.

Science Feedback is a non-partisan, non-profit organization dedicated to science education. Our reviews are crowdsourced directly from a community of scientists with relevant expertise. We strive to explain whether and why information is or is not consistent with the science and to help readers know which news to trust.
Please get in touch if you have any comment or think there is an important claim or article that would need to be reviewed.

Published on:

Related Articles