How Belarusian State Media Talked About Its EU Countries: August 2026

Key Findings / TL;DR

• Top target by volume: 🇫🇷 France: 2,013 materials. Followed by: 🇵🇱 Poland (1,892), 🇱🇻 Latvia (1,686).

• Top narrative: Internal problems: 753 materials in deep analysis (25% of identified propaganda).

• Multimodal FIMI on YouTube: Baltic cluster 27.7%, Weimar Triangle 29.8%. Highest among countries with 20+ videos: 🇵🇱 Poland (39.4% on 104 videos).

• AI: 3,667 materials processed, 1470 flagged as HIGH risk (40.1%). 2982 identified as propaganda (excluding neutral and not-about-targets).

• Coordination: TASS cited in 136 materials, the most cited external source. 130 coordination events (low confidence, see below) across 31 days.

🔶 Weimar Triangle: Two Attack Strategies

The Weimar Triangle countries (Poland, Germany, France) receive fundamentally different propaganda treatment than the Baltic states. While the Baltics face a monotonal military threat narrative, each Weimar country is attacked through its unique perceived vulnerability.

🔵 Baltic States: military monotone Military buildup 29% Internal problems 22% Border tensions 17% Economic decline 9% Migration 7%

🔶 Weimar Triangle: diversified attack Internal problems 28% Military buildup 18% Migration 14% Economic decline 10% Arms to Ukraine 9%

🟢 Visegrád Group (V4): the Central European frame

The Visegrád Four (Poland, Czech Republic, Hungary, Slovakia) are framed through a third lens. This month the cluster’s leading category is “Internal problems” (27%). Poland appears in both Weimar and V4 clusters as an analytical lens, not double-counted in totals.

🟢 V4: “Internal problems” leads Internal problems 27% Migration 18% Military buildup 18% Arms to Ukraine 9% Historical revisionism 7%

🇵🇱 Poland: two contexts In the Weimar context Poland is attacked as a Western military partner; in the V4 context it appears among Central European neighbours where the EU-division frame dominates. One dataset, two strategic constructions.

🇱🇻 Latvia: the month’s main shift

Latvia accounts for 1,686 materials in August, the largest increase in this issue on a source base comparable with July: up 50%, while Poland and France stayed at their July level and Germany fell by a quarter. The rise does not come from a single outlet: national news sites, YouTube and Telegram all grew, and the level held throughout the month, peaking in the first week of August. Six in ten Latvia materials now come from Telegram.

There was one point of entry: the temporary closure of a crossing point on the Latvian-Belarusian border in the first days of August. According to the same state media, the Latvian government did not back the initiative and traffic resumed on the evening of 4 August. Yet a brief episode became the frame for the whole month. The Belarusian Foreign Ministry said that by closing the border Latvia was trying to ‘pump money out of the European Union’ and called the decision ‘a cynical pre-election step’; ONT carried the version that it had been ‘a planned sounding-out of the electorate’, and SB wrote about EU money and fear of the visa-free regime. Vitebskiye Vesti, the paper of the only Belarusian region bordering Latvia, told the same story as a neighbour’s retreat: ‘they initiated closing the border with Belarus but then backed down’. Latvia is that paper’s leading EU country for the month.

Familiar lines were built around this core: militarisation of the border (‘Mining their own future’), the neighbour’s supposed non-viability (‘Can Latvia survive on its own?’) and a contrast between Belarusian and neighbouring authorities (‘Belarusians can rely on their government, while residents of Lithuania and Latvia have no such luxury’). The high-risk share for Latvia is 41% (25% by both models), below Poland (54%) and Lithuania (47%). The Saeima election on 3 October is mentioned in only 27 Latvia materials in August, but already within a pre-election frame: candidates are called ‘open Nazis’, and pro-government Telegram gives a platform to a former Saeima deputy explaining what Latvia should learn from Belarus. This is the baseline against which September will be read.

🇭🇺 Hungary: a fourth month without a personal line

270 materials in August; on thumbnails Orbán 0, Magyar 0 (propaganda 6.8/10). The personalised Hungarian storyline, closed since May, has not returned, but the texts show a nuance: Magyar is named in the headlines of nine materials, and all of them are news-agency items rather than attacks: the election of András Baka as president, Hungary meeting EU conditions for €10 billion, doubts over the environmental permit for the Paks II nuclear plant. The high-risk share is 21% (12% by both models), one of the lowest in the issue, and the month’s leading category is neutral news (41 of 119 materials analysed). In August Hungary is not a target for Belarusian state media but a country from the agency wire.

🟠 Iberia (ES, PT)

Spain and Portugal have been monitored since the July 2026 issue. Volumes remain an order of magnitude below the core clusters (775 materials on Spain, 96 on Portugal); the cluster is read mainly through Spain, with Portugal as background.

🟠 Iberia: narrative mix Migration 66% Internal problems 12% EU internal division 7% Border tensions 5% Economic decline 5%

🇪🇸 Spain Migration 70% Internal problems 10% Border tensions 6% Economic decline 5% EU internal division 5% 🇵🇹 Portugal EU internal division 32% Internal problems 26% Migration 21% economic_war 9% Western aggression 3%

Ceuta and the tunnels: someone else’s crisis versus one’s own

Spain accounts for 775 materials in August. On a source base comparable with July this is a rise of about a quarter, and almost all of it is Ceuta. The migration crisis in the Spanish enclave, which state media picked up in the last days of July, became the main foreign story of the first week of August: 468 materials over the month, almost half of them (226) on 1–6 August, a second surge on 14–19 August (93) and a fade to single mentions after the 20th. The story was carried by Telegram channels (262 materials) and the websites of national state media (187, led by SB, CTV and BelTA with 76, 59 and 47); regional newspapers barely noticed it (4). The model assigned more than half of the Spain materials analysed (265 of 466) to the ‘migration as a weapon’ category.

The frame was set from day one: not a humanitarian situation on the border with Morocco but a ‘migrant tsunami’ that split Europe. Around agency reports on migrants returning to Morocco, pro-government Telegram also spread conspiracy versions, for instance that the migrant ‘invasion’ of Spain was organised by Morocco with the backing of the US and Israel. YouTube reduced the story to the formula ‘Brussels in panic’. Portugal (96 materials) remained background with scattered stories; its link to Ceuta is, for instance, tighter control of its maritime border. FIMI indicators appear on 24.5% of the cluster’s thumbnails, but the video base is small and the share is not carried into the conclusions.

In the same weeks a migration story of Belarus’s own was unfolding on its border. On 3 August the Zapadny Rubezh channel reported that Lithuania accused the Belarusian authorities of helping migrants build tunnels under the border; on 17 August BelTA reported that the head of Lithuania’s border guard service had resigned after an incident on the border. State media devoted about 35 materials to tunnels under the Lithuanian and Latvian borders, roughly thirteen times fewer than to Ceuta. The register ranges from mockery to attempts at analysis. On one side is denial: ‘An invented threat’ in SB, ‘myths about tunnels’ on YouTube, a ‘funny tale’ told by Lithuanian border guards on the ZhS Premium channel, forwarded almost seven hours later by Myasorubka, or ‘supposedly 12 tunnels’ on the Nevolfovich channel. On the other are texts that acknowledge the tunnel but turn the conversation to a Lithuanian border service supposedly hiding incidents, and from there to whether claims of a Belarusian ‘hybrid war’ can be trusted. Tellingly, one such post in a pro-government channel turned out to be a reprint from Teni Pribaltiki, a pro-Kremlin Baltic channel aimed at Russian-speaking audiences in the Baltic states. At the edges of the spectrum the accusations harden: the head of Lithuania’s border service is accused, without any source, of ‘schemes with migrants’ and cigarette smuggling, and the Nevolfovich channel compares Lithuanian border guards to the collaborators of 1941 in a photo collage. In both registers responsibility for the border lies with Lithuania, and the accusation against the Belarusian authorities is not addressed on its merits.

Someone else’s crisis is presented as proof of Europe’s weakness; one’s own, on one’s own border, as the neighbours’ invention or failure. This is the agenda substitution we promised to examine in the July issue: the migration theme does not disappear, it is moved to where Belarus looks like a spectator rather than a participant.

🟣 Benelux (NL, BE, LU)

The Netherlands, Belgium and Luxembourg have been monitored since the July 2026 issue (173, 114 and 6 materials). The cluster is read through the Netherlands; Belgium is background and Luxembourg is kept as a footnote without a country profile. Capitals are deliberately excluded from country matching: in this corpus “Brussels” means the EU and NATO in over 90% of cases and “The Hague” means international courts, not the country.

🟣 Benelux: narrative mix Internal problems 24% EU internal division 18% Military buildup 12% Economic decline 12% Arms to Ukraine 7%

🇳🇱 Netherlands Internal problems 24% EU internal division 16% Military buildup 14% Economic decline 14% Arms to Ukraine 10% 🇧🇪 Belgium Internal problems 25% EU internal division 21% economic_war 14% Economic decline 11% Migration 11%

* Luxembourg is counted in the Benelux total but has no country profile: 6 materials this month.

Benelux: Belgium as an argument for EU division

In the second month of monitoring the cluster (293 materials) again has one story of its own, again Belgian (114), but in a different role. In July Belgium was the militarisation exhibit; in August it is the country that rejected using frozen Russian assets to finance Kyiv; pro-government Telegram presented this as ‘the door is closed’, adding that the others are afraid of responsibility. Once again Belgium is not a country in its own right but an argument, this time for the thesis of a divided EU. A second motif, the Brussels prosecutor admitting police powerlessness against street gangs, feeds the familiar internal-problems frame. The Netherlands (173) is examined separately below; Luxembourg (6) appears only as a footnote. The cluster’s thumbnail FIMI share is 32.1%, on a small base of omnibus shows, so it is not carried into the conclusions.

🇳🇱 Netherlands spotlight

173 materials, 16 videos mentioning the country, FIMI indicators on 25.0% of thumbnails. In August the Netherlands for the first time has a story of its own, if a small one: farmers’ protests against environmental policy and ‘nitrogen restrictions’. It was relayed both by agencies (BelTA) and by pro-government Telegram, where the picture matters more than the text: a formation of more than 500 tractors, thousands of tractors on main roads, ‘tractors back on the streets’. This continues the July line, when farmers appeared in the headline of an SB TV talk show as an illustration of a ‘revolt against the EU’, but in August the story stood on its own.

Otherwise the picture is unchanged: the Netherlands mostly appears in lists, in the themes of arms deliveries to Ukraine, EU division and ‘Europe preparing for war’. The high-risk share is 23% (13% by both models), one of the lowest in the issue, and the leading category is neutral news (26 of 79 materials analysed). For Belarusian state propaganda the Netherlands is still not a target but part of the background called ‘Europe’, and that is what makes the farmers’ story interesting: it is how a country otherwise barely noticed enters the agenda.

📺🖼️ YouTube: Multimodal Analysis by Target Country

The text layer of the mention analysis (threat categories per video) combined with thumbnail visual analysis (Vision LLM, codebook v2.1) reveals how the same propaganda system uses different visual packaging for different target countries.

Cluster comparison

Cluster	Videos	Clickbait	Propaganda	FIMI %*
🔵 Baltic States	184	3.21	6.0	27.7%
🔶 Weimar Triangle	349	3.25	6.15	29.8%
🟢 V4	186	3.32	6.02	32.8%
🟠 Iberia	53	3.32	6.02	24.5%
🟣 Benelux	28	3.47	6.71	32.1%

Country	Videos	Clickbait	Propaganda	FIMI %*	Top visual technique
🔶 🇵🇱 Poland	104	3.42	6.08	39.4%	appeal_to_authority
🔵 🇱🇹 Lithuania	66	3.41	6.1	36.4%	appeal_to_fear
🔵 🇱🇻 Latvia	95	3.05	5.82	22.1%	appeal_to_authority
🔵 🇪🇪 Estonia	23	3.3	6.48	26.1%	appeal_to_fear
🔶 🇩🇪 Germany	104	3.39	6.23	31.7%	appeal_to_authority
🔶 🇫🇷 France	141	3.02	6.14	21.3%	appeal_to_authority
🟢 🇨🇿 Czech Republic	47	3.06	5.28	19.1%	appeal_to_authority
🟢 🇭🇺 Hungary	24	3.5	6.8	29.2%	appeal_to_fear
🟢 🇸🇰 Slovakia	11	3.09	6.89	36.4%	appeal_to_authority
🟠 🇪🇸 Spain	49	3.31	6.27	22.4%	appeal_to_fear
🟠 🇵🇹 Portugal	4	3.5	2.95	50.0%	appeal_to_fear
🟣 🇳🇱 Netherlands	16	3.38	6.88	25.0%	appeal_to_fear
🟣 🇧🇪 Belgium	11	3.73	6.57	45.5%	demonization

* FIMI indicator under visual codebook v2.1 (EEAS operationalisation). Confidence in the visual layer is moderate; see the section on confidence in findings.

🧭 DISARM: Tactics & Techniques (indicative)

Since the June 2026 issue, narrative categories and manipulation types identified by the AI analysis are translated into techniques of the open DISARM Red framework — the shared vocabulary of EEAS, FIMI-ISAC and the counter-disinformation community. Counts are materials carrying indicators of a technique; one material can map to several. Deterministic v0 translation layer, not per-item behavioural coding.

ID	Technique	Materials
T0135	Undermine	1059
T0023	Distort Facts	912
T0078	Dismay	803
T0075.001	Discredit Credible Sources	753
T0079	Divide	730

Coordination layer: T0002: 255, T0084: 255, T0119: 62.

High-propaganda thumbnails: what the audience sees first

Real YouTube thumbnails from videos scored 8–10/10 on propaganda scale. These images are the first contact point with the audience — designed to trigger emotional response before the content is even consumed.

NEWS.BY Поляки ненавидят украинцев | Европейский нацизм | Россия уни… 10.0/10Estonia, France, Germany	СБ ТВ 🔴 ТОЧКИ НАДРЫВА: Иран наступает, США теряют контроль, а где … 10.0/10Czech Republic, Poland	СБ ТВ 🔴 США без ракет, Европа в тупике! Риски глобального конфликт… 10.0/10Belgium, France, Spain
СБ ТВ 🔴 Провокации на границе Беларуси, иранская боль США, украинс… 10.0/10Czech Republic, France, Germany	СБ ТВ 🔴 Польский реванш и дно украинской ямы: какой выбор у Белару… 10.0/10France, Germany, Latvia	NEWS.BY Что скрывает День Войска Польского? | Адаптация КНДР и имидж… 10.0/10France, Germany, Poland
NEWS.BY Европа теряет весь урожай? | Секрет успеха Поднебесной | "МЕ… 10.0/10Czech Republic, France, Germany	СБ ТВ 10 ПОЗОРНЫХ ФАКТОВ О ПОЛЬШЕ до 1939: диктатура, концлагеря и… 10.0/10France, Germany, Poland	NEWS.BY Когда ждать Кушнера и Уиткоффа в Москве? | Подстава для "ПиС… 10.0/10Czech Republic, France, Germany

Cross-modal divergence: thumbnail vs content

A distinct May finding: for a share of videos the thumbnail is alarmist while the transcript’s assessed threat is low. The cover does emotional work the content does not. Divergence is highest for 🇵🇹 Portugal (75.0%).

Country	Divergence	Alarmist cover / low-threat text
🟠 🇵🇹 Portugal	75.0%	3/4
🟣 🇧🇪 Belgium	45.5%	5/11
🟣 🇳🇱 Netherlands	31.2%	5/16
🔶 🇵🇱 Poland	30.8%	32/104
🔵 🇱🇹 Lithuania	28.8%	19/66
🔶 🇩🇪 Germany	27.9%	29/104
🟢 🇭🇺 Hungary	25.0%	6/24
🟢 🇨🇿 Czech Republic	23.4%	11/47
🔵 🇱🇻 Latvia	21.1%	20/95
🟢 🇸🇰 Slovakia	18.2%	2/11
🔵 🇪🇪 Estonia	17.4%	4/23
🟠 🇪🇸 Spain	16.3%	8/49
🔶 🇫🇷 France	12.8%	18/141

CHECKGermany Главным фактором высоких цен на электроэнергию в Германии является тяжелое бремя налогов и сборов ↗

CHECKGermany Канцлер ФРГ Фридрих Мерц лишая граждан социальных льгот. ↗

CHECKGermany «Какой прок от самого прекрасного пособия по уходу за ребёнком, если мир в Европе под угрозой?» — сказал Мерц. ↗

CHECKGermany Канцлер Германии Фридрих Мерц косвенно поставил расходы на оборону выше родительских пособий ↗

CHECKGermany Он намерен сделать оборонные проекты по модернизации Бундесвера «гораздо более экономически эффективными» ↗

CHECKGermany Закрытие немецких атомных электростанций … стало одной из причин высоких цен на электроэнергию в стране ↗

CHECKGermany По данным портала по сравнению тарифов Verivox, Германия заняла первое место по уровню цен на электроэнергию среди стран G20 во вт ↗

CHECKGermany Средняя цена киловатт-часа для немецких домохозяйств более чем в два раза дороже, чем средний показатель по G20 ↗

CHECKLithuania Большинство путешественников приезжали из соседних стран – Литвы, Латвии и Польши ↗

CHECKNATO/EU Всего за время действия безвиза нашу страну посетили 1 422 851 житель Европы ↗

Conclusions

1. Two cluster strategies persist. In August 2026 the Baltic cluster (LT, LV, EE) accumulated 3,164 mentions, while the Weimar Triangle (PL, DE, FR) reached 5,471 mentions. Multimodal YouTube analysis shows Baltic FIMI at 27.7% and Weimar at 29.8% — both clusters operate as fully-formed propaganda targets, with Weimar slightly more intense.

2. 🇫🇷 France: top target by volume. With 2,013 materials, this country dominates the monitoring period. The dominant narrative is 38% Internal problems.

3. YouTube as propaganda amplifier. Multimodal analysis covered 696 video–country pairs across the target countries. Among countries with 20+ videos, the largest share of FIMI-flagged thumbnails is in content about 🇵🇱 Poland (39.4%).

4. TASS as the main external source. TASS cited in 136 of analyzed materials, followed by RIA Novosti (83). This shows reliance on Russian state agencies; it is not in itself evidence of coordination, which this series rates as low confidence.

5. Factual manipulation dominates. Of all detected manipulation types, factual manipulation (cherry-picking real data to distort conclusions) accounts for 4308 cases — more than emotional (2948) and logical (2319) combined. This makes the propaganda harder to detect than outright fabrication.

6. Iberia and Benelux remain an order of magnitude smaller. Iberia (ES, PT) accumulated 871 materials and Benelux (NL, BE, LU) 293, against 3,164 for the Baltic cluster and 5,471 for Weimar. Each of the two clusters is read through one country (Spain, the Netherlands); the rest is background. Country matching uses country stems and demonyms only; capitals and politicians were tested and excluded because most of their mentions do not concern the country.