IN THIS EDITION
ON THE RISE
Weekly Watch of Emerging Disinformation Risks
War in Ukraine: AI-powered disinformation targets mobilization
Disinformation about the war in Ukraine has been a constant phenomenon since Russia launched its full-scale invasion in February 2022. In recent months, according to EDMO monitoring, it has consistently been the main target of disinformation in the European Union, alongside migration.
Russian-driven disinformation narratives are recurrent: the Ukrainian government, military and population are largely pro-Nazi; Ukrainian refugees are violent or parasitic, and European countries treat them better than their own citizens. Zelenskyy and his entourage are drug users and corrupt. The war is NATO’s fault. It is the West that is seeking escalation. And so on.
A particularly rich strand, and one typical of any conflict, is of course war propaganda and disinformation. Its function is straightforward: to bolster the morale of one’s own side and undermine that of the adversary. It has existed for as long as wars have, and it is used by both sides in a conflict.
A significant development over the past two years, however, has been the growing use of Artificial Intelligence to create short videos — usually between 5 and 15 seconds long — that are realistic enough to appear credible, even though they are often not technically deepfakes. These videos depict precisely what propaganda would like to be true. A significant wave had already been detected by EDMO in late 2025, in connection with the Russian offensive on Pokrovsk, with numerous AI-generated videos spread by pro-Russian accounts showing Ukrainian soldiers surrendering or crying.
In September, a new wave of pro-Russian AI-generated videos was detected, this time focusing in particular on the mobilization of Ukrainian soldiers to the front. Several videos and images show soldiers in handcuffs on buses, being forcibly taken away. This is happening, perhaps not coincidentally, alongside some successes by the Ukrainian armed forces in the Donbas area — in particular the liberation of several towns north of Lyman, as part of Operation Vivaldi — and as reports intensify of a possible new mobilisation in Russia.
The message conveyed by Russian disinformation can potentially target several audiences: the Russian public, the Ukrainian public, and public opinion in European countries allied with Kyiv. Not coincidentally, these videos are being recirculated in several European countries with subtitles in different languages. And AI-generated videos and images are not the only tools being used. The campaign is also reinforced through the “old” technique of taking real videos or images and distorting their meaning with false captions. For example, videos and images of a protest against the cutting down of trees in Kyiv have been circulated by pro-Russian disinformation with the false claim that they showed Ukrainian civilians being taken away by soldiers because they were trying to disrupt a recruitment operation.
The most concerning element is the relatively low quality and limited coordination of these propaganda and disinformation operations, which combine generative AI with other techniques. The potential for technical and tactical improvement is enormous. If this type of content is already capable of causing harm by being believed to be genuine by more vulnerable segments of the public, thinking about what it could do in the future is chilling.
ZOOM-IN
A Closer Look at Cases Detected by the EDMO Network
The war in Ukraine is increasingly fought in the digital information space, where AI-enabled tools amplify deceptive narratives, manipulate images, and facilitate targeted recruitment. From fabricated job opportunities used to lure foreign fighters for recruitment in the Russian army, to false claims of mass flight from Kyiv and AI-generated “evidence” about military equipment, The following articles collectively highlight how disinformation exploits uncertainty and fear to influence mobilization and perceptions of the war.
Russia recruits Africans for the war in Ukraine with fake job ads on social media
Russia is using deceptive social media ads – mainly on Facebook – to recruit young men from Africa, the Middle East, Asia, and Latin America for its war in Ukraine. A report by the NGO Code for Africa referenced by TjekDet exposed 33 recruitment groups running nearly 300 websites and social media accounts. These groups post paid Facebook ads – often featuring AI-generated images – promising lucrative jobs as security guards, actors, or drivers in Russia with monthly salaries exceeding $3,000. Upon arrival, the men are pressured into signing Russian-language military contracts and sent to the frontlines. Many face confiscated passports, threats, or unpaid bonuses; one in five identified African recruits has been killed. Experts from the Danish Institute for International Studies note this foreign recruitment surge offsets domestic soldier shortages while shielding Russia’s middle class from mobilization.
Ironically, these campaigns rely heavily on Western servers and software tools, including:
- Hosting Infrastructure: the recruitment group known as “IRC Russian Bear” hosts and distributes its digital campaign materials from a server physically located in Amsterdam, Netherlands;
- AI Creation Software: multiple recruitment landing pages and highly persuasive campaign graphics are constructed using artificial intelligence generation tools developed by companies in the United States and Sweden.
- Data Collection Software: when interested applicants click the deceptive Facebook ads, their private contact details, background profiles, and passport information are collected using software owned by an Austrian tech company before being forwarded directly to operators in Russia.
False ‘mass exodus’ claims spread as Russia bombards Kyiv
The article debunks claims that Kyiv is experiencing a mass exodus as Russia intensifies its aerial attacks on the city. Social media posts have circulated photos and videos allegedly showing residents fleeing, but fact-checkers found that much of the material was misleading or unrelated. One video showed ordinary rush-hour activity in a Kyiv metro station, while an image supposedly depicting a traffic jam leaving Kyiv was actually taken at a toll station in China. Another photograph showed Ukrainians moving toward the Polish border in 2022, not residents fleeing Kyiv in 2026. A genuine video of traffic in Kyiv was also misrepresented as evidence of people escaping the city. Analysts say these narratives exaggerate real hardships to create panic, undermine morale and potentially weaken international support for Ukraine. There is no public evidence of a current mass departure.
AI models say the sea drone destroyed near Neptun Deep was “undoubtedly” Ukrainian
AI-generated assessments were used to amplify a false claim that a maritime drone destroyed near Romania’s Neptun Deep gas platform was “undoubtedly” Ukrainian. Several Romanian media outlets, Facebook pages and Telegram channels cited supposed analyses by ChatGPT, Gemini and Grok, presenting them as authoritative evidence. The narrative gained significant traction online: one article received over 1,000 reactions, 240 comments and 300 shares on Facebook, while another generated more than 3,000 reactions, 700 comments and 800 shares.
However, Fact Hub’s experiment found that when given identical photographs and neutral questions, the three AI models produced substantially different identifications, ranging from a possible British-derived vessel to a likely Russian-origindrone and a Ukrainian MAGURA V5. The article therefore demonstrates how AI outputs can acquire credibility and resonance on social media despite uncertainty, contradictory evidence and the models’ own acknowledged limitations.
ELECTION BEAT
Tracking electoral disinformation through EDMO Hubs
Sweden’s election aftermath: How misinformation shaped the campaign
Sweden’s recent elections saw misinformation emerge around voter fraud, electoral interference and AI-generated content. Fact-checkers from Källkritikbyrån, NORDIS partner, and researchers at the Fojo Media Institute monitored viral claims throughout the campaign, finding that minor voting irregularities and discarded ballots were often presented as evidence of deliberate fraud. AI-generated images also contributed to misleading narratives, including a fabricated story showing ballots dumped into a river.
Researchers additionally identified conspiracy theories alleging manipulation by “globalist elites” and portraying immigrants as being used to influence the vote (like this AI-generated picture spreading Islamophobic disinformation), narratives that continued circulating after election day. While foreign actors were active in amplifying misleading content, researchers found no immediate evidence of large-scale coordinated foreign interference or significant impact on voters. Instead, much of the misinformation appeared to originate domestically, highlighting how existing political narratives, conspiracy theories and readily available AI tools can contribute to information disorder around elections.
Postal voting as a target of disinformation campaign at MP elections in Berlin and Mecklenburg-Vorpommern
A staged video circulated online during Berlin’s September 2026 parliamentary election, falsely claiming to show postal ballots for the AfD being deliberately destroyed while CDU votes were preserved. The footage fuelled accusations of election fraud, but AFP (GADMO partner) found that the envelopes and documents shown differed from official Berlin postal voting materials. The Berlin state election authority also confirmed that the video was fake and highlighted inconsistencies with official vote-counting procedures.
The video was part of a wider pattern of false claims surrounding postal voting during German state elections in autumn 2026. Similar content alleging manipulation in favour of or against particular parties circulated in Mecklenburg-Vorpommern and Saxony-Anhalt. Some posts attracted unusually high numbers of views compared with their engagement, suggesting possible artificial amplification. Several accounts spreading the claims were also linked to pro-Russian disinformation networks previously associated with campaigns targeting European elections and promoting narratives favourable to Russia and the AfD.
Vučić and snap elections: has the President been inconsistent?
Serbian President Aleksandar Vučić announced extraordinary parliamentary elections for 25 October 2026, despite previously linking the timing of an early vote to the international security situation. In April, amid calls from protesting students for early elections, Vučić said there would be no summer vote if a major global crisis persisted, suggesting elections could instead take place if progress towards peace was made.
Over the following months, Vučić continued to warn about the economic and security consequences of the wars in Ukraine and the Middle East. In early September, he expressed concern about growing international tensions and potential wider conflicts, while predicting that the situation would worsen during the winter. Nevertheless, on 9 September, following a proposal from the Serbian government, he dissolved parliament and called an early election. Serbian fact-checking outlet Istinomer therefore assessed his position as inconsistent with his earlier statements linking elections to an improvement in the international situation.
GLOBAL PULSE
Disinformation narratives shaping the world’s conversations
Where business, politics, and propaganda join hands. Who is creating climate disinformation?
Climate disinformation emerges from the intersection of business interests, political actors, propaganda and influence operations. Demagog.pl argues that contemporary climate disinformation rarely denies climate change outright. Instead, it more often presents climate policies as ineffective, unrealistic, unfair or economically damaging. The article identifies fossil-fuel interests, political actors, think tanks, media, influencers, bots and trolls as participants in the broader disinformation ecosystem. It highlights documented historical cases involving companies such as Exxon and Shell, whose internal documents acknowledged climate risks while public communications emphasized scientific uncertainty.
The article also examines Russia’s role, arguing that climate-related narratives can serve Moscow’s economic and geopolitical interests by weakening support for decarbonisation and exploiting divisions over energy prices, renewables and EU policies. Many anticlimactic narratives arise locally and have local economic or political roots. Russia often doesn’t need to invent conflicts – it simply seeks out existing tensions and amplifies them on social media. A similar mechanism can be observed in times of crisis and natural disasters. For example, following the tragic flooding in Valencia, Russian messages reinforced existing criticism of the Spanish authorities and linked the country’s weakness to support for Ukraine.
Do solar panels really raise the temperature?
The article examines claims that solar panels significantly increase temperatures and therefore undermine their climate benefits. It finds that photovoltaic installations can indeed alter local temperatures, but the effect varies considerably depending on location, weather, panel efficiency and what surfaces they replace. Studies cited by Les Surligneurs have recorded increases of up to 5.2°C near rooftop panels during extreme heat, while other research has found nighttime warming of 3-4°C above ground-based solar farms. Conversely, panels can sometimes reduce temperatures, particularly when replacing heat-absorbing surfaces such as asphalt, and some studies have found cooler nighttime conditions. The article stresses that these localized effects should not be confused with global climate impact. Over their life cycle, solar panels produce substantially fewer greenhouse-gas emissions than fossil-fuel electricity. Although estimates of their overall climate benefits vary, the available evidence indicates that photovoltaic energy has a considerably lower carbon footprint than fossil fuels.
Sharepic cites ten-year-old figures on alleged development aid to China
A social-media graphic claims that Germany gives China €630 million in annual development aid while China builds hundreds of coal-fired power plants and plans hundreds more. GADMO finds the claim misleading because it combines outdated figures with current political debates. The €630 million figure dates from 2017 and referred largely to repayable loans, not conventional development aid. Germany has since ended its classic bilateral development cooperation with China. Current German-funded projects in China are much smaller, with around €19 million committed to environmental and social initiatives, and the money is not paid directly to the Chinese government. The figures pointing to about 368 coal plants under construction and 803 planned are also around a decade old. However, the article notes that China continues to expand coal power: 94.5 GW of new coal projects began construction in 2024. At the same time, China is rapidly expanding solar energy, which by July 2026 had slightly exceeded coal in installed capacity.
ON A DIFFERENT NOTE
Lack of platform accountability was the most widely shared concern, rated a big or extreme threat by 78% of experts, narrowly ahead of misinformation, 73%, and polarization, 69%. The dominance of a few social media platforms, 64%, and filter bubbles or echo chambers, 53%, followed. Generative AI was rated the smallest of the six threats, though still a big or extreme threat for 47% of experts. […] Finally, experts reported the barriers they face in their work on the information environment. Availability of research funding was the most common barrier at 67%, ahead of data access at 60%. Funding overtook data access as the top barrier in 2025 and remained the top barrier this year. […] Pressure to align research with a government’s agenda rose from 21% to 33%, government censorship from 11% to 23%, and pressure from private companies from 8% to 18%.
International Panel on the Information Environment “Trends in the Information Environment: 2026 Expert Survey Results“
Paolo Cesarini, Editorial Director
Tommaso Canetta, Editor-in Chief
Editorial Staff include Elena Coden, Paula Gori, Elena Maggi
This edition draws in part on automated translation and reflects information available as of 30 September 2026. Later developments may not be included.
