On 16 July, Communications Minister Kirill Zalessky signed, on behalf of Belarus, the agreement founding the World Artificial Intelligence Cooperation Organization (WAICO). Twenty-nine states signed in Shanghai the day before the World AI Conference opened; the body is headquartered there; Beijing proposed it a year ago and now anchors it. The other signatures tell you what the room was: Russia, Cuba, Venezuela, Serbia, Kazakhstan, Pakistan, Indonesia, Brazil, plus a dozen or so African and Asian states. No United States, no EU member, no India.

By any ordinary standard this is a substantial move. Belarus is now a founding member of an intergovernmental organization meant to write technical standards and governance principles for AI, with a permanent seat at a table Beijing is building as an alternative to Western AI governance. Zalessky’s own comment to BelTA put it plainly: founding membership “opens up broad opportunities to help shape the international agenda.”

We went looking for how that landed at home. It didn’t.

Thirteen documents

We searched the full FORESIGHT corpus, about three million documents from 73-plus sources, for the July window around the signing. WAICO, WAIC as a standalone word, the Russian name of the conference. Everything.

Thirteen documents. Five of them are posts on the ministry’s own Telegram channel. Two are BelTA wire items. Two came from pro-regime Telegram, one from Sputnik Belarus, one from the business outlet PrimePress. That leaves two documents from the entire democratic media space: one item from Pozirk, and one post on an independent Telegram channel, four days later.

Of the eleven independent outlets we track weekly — Zerkalo, Nasha Niva, Reform, Belsat, Radyjo Svaboda, Euroradio, Mediazona, Gazeta.by, Belmarket, Plan B, Pozirk — ten published nothing at all.

Before calling that a finding, we checked the obvious objection. Maybe independent outlets simply don’t cover AI. They do: in July they published 159 documents on AI topics, against 189 from state media. The beat exists, it is roughly the same size as the state’s, and none of it was pointed at Shanghai.

Maybe, then, they ignore ministry announcements and foreign-policy set pieces generally. They don’t. Take the other big state initiative of the same fortnight, Lukashenko’s invitation to Uzbek labour migrants: 286 independent documents against 88 from state media. On that story independents out-covered the state more than three to one.

So the silence is specific. It is not a missing beat and not a general aversion to official news.

The state barely pushed it either

The more surprising half of the result is on the other side. Nine of the thirteen documents are state or pro-regime, but nine documents is nothing. A founding membership in a new international organization generated two BelTA items and a handful of ministry Telegram posts, then stopped. Slavianski Bazaar, running the same week, generated coverage in a different order of magnitude.

Whatever this membership is for, domestic legitimation does not appear to be it. The Shanghai messaging was pitched outward: at Beijing, at the other twenty-eight signatories, at the Global South framing Xi used in his own address. The ministry’s statement leans on technological sovereignty, on the “all-weather comprehensive strategic partnership” with China, on readiness to cooperate through the Union State, the EAEU, the CIS and the SCO. That vocabulary sits on the ministry’s website. It did not travel: the three state documents in our sample that carry the story mention China and Xi, and essentially nothing else from that list. With three documents we can’t measure propagation properly — but there is nothing to measure, which is the point.

One related number, since it bears on where Belarusian audiences meet Chinese framing at all: Xinhua was cited in 30 state-media documents in July and 2 independent ones.

What the citizens think, since nobody asked them

Zalessky told the high-level meeting that the goal is to give “every industry of the economy, every state body, every citizen access to the most modern capabilities of artificial intelligence.”

In March, FactCheck.LT ran a synthetic survey on AI attitudes: 1,500 personas built from 2.47 million real comments across YouTube, TikTok and Telegram, cross-checked on two frontier models. It remains the only systematic measurement of AI attitudes in a country where ordinary polling on anything sensitive is impossible.

61.6% had never tried an AI tool. Another 21.4% tried one and dropped it. Regular users, monthly or more, came to 8.2%. Age does nearly all the work: a quarter of 18-to-24-year-olds use AI monthly, against 92% of the over-55s who have never touched it. Political orientation, which splits this population on almost everything else, barely moves the number.

A caveat that cuts against our own headline: most people who have met AI met it inside something else, translation apps and voice assistants, not chatbots. Only 8.3% had tried ChatGPT, Claude or Gemini directly. Some of the “never tried” group uses AI daily without filing it under that name.

Then the finding that makes this more than a gap story. Asked whether the state should regulate AI, 96.3% said yes: 50.2% strict regulation, 46.1% moderate, 3.5% opposed. Nothing else in our survey programme comes back this stable — three independent runs within two percentage points, cross-model agreement within 1.2, and every political segment converging. Ask these same respondents about an election and they scatter. Ask about AI rules and they agree.

Ask what AI should be for and the answer is equally lopsided: healthcare, 88.1%, far ahead of manufacturing at 60.5% and education at 42.1%. Public administration, the priority Zalessky named second in Shanghai, is not where this population is looking.

Two agendas that never meet

Put the three pieces together and you get a policy with no domestic audience.

The state’s AI agenda is infrastructural and geopolitical: sovereign data centres, locally hosted models, competence centres, a seat in a China-led body, cooperation offered through the Union State, the EAEU, the CIS and the SCO. The public’s version, as far as 1,500 synthetic personas can stand in for it, is about protection and hospitals: rules first, healthcare first, and almost no expectation that any of it touches their own job — only 2.7% see AI as a serious professional threat within five years, against 47% expressing general anxiety.

And between the two, no transmission belt. The state did not campaign on Shanghai; the independent press did not contest it. A founding membership in the organization that intends to write global AI standards entered Belarusian public discourse as thirteen documents, and left.

That is worth noting for a practical reason. AI governance is one of the very few policy areas where this population agrees with itself across the political divide. If the Belarusian public ever gets a debate about who regulates AI and how, the ground for it is unusually solid. So far, the debate has not started, and neither side of the media system has tried to start it.

Corpus method: search of the full FORESIGHT corpus 3.3M documents, 100+ sources) for 10–31 July 2026 on WAICO, word-boundary WAIC, and Russian-language names of the organization and conference; the AI-beat and Uzbek-labour comparisons cover 1–31 July and 9–19 July respectively. Counts are documents, not articles: syndicated wire copy can appear more than once. Survey: FactCheck.LT synthetic survey of AI attitudes, March 2026, N=1,500 personas from 2.47M comments, cross-validated on Claude Sonnet 4 and GPT-5-mini; three-run mean absolute deviation on the regulation question 1.9pp. The sample is active social media commenters and overstates digital engagement relative to the general population; in this method attitudinal findings are more reliable than behavioural ones. Other sources: Ministry of Communications and Informatization statement, 20 July; BelTA, 17 July; Reuters, Xinhua and Global Times on the signatories.

The official half-year number came out in July: GDP up 1.5%, against an annual plan of 2.8%. State outlets wrapped the release in the usual vocabulary of stability. Independent ones did the arithmetic and led with the miss. Worth remembering that 2.8% was already a climbdown; earlier programme documents talked about growth of “at least 4%.”

So far, an ordinary story about an economy underperforming its plan. It gets less ordinary when you ask where the fiscal consequences of that underperformance would normally show up.

We spent this year turning Belstat’s paper archive into a database. 988 publications, 1995 to 2026, just under six million extracted observations, sitting next to the FORESIGHT discourse corpus of about three million media documents. One thing this makes possible is a page-by-page comparison of the same yearbook across editions. We ran it on thirteen editions of “Belarus in Figures,” the state’s flagship annual digest.

The government finance chapter (taxes, budget, public debt) appears in every edition from 2014 through 2022. From the 2023 edition on, it is gone. The energy section lasted two editions, 2021 and 2022, then disappeared the same year. Social protection dropped out earlier. The diff also caught 2,096 cases where a previously published value was quietly changed in a later edition.

Note the year. The 2023 edition was the first one compiled entirely after the 2022 sanctions waves. Belstat obviously still counts the budget; the finance ministry could not function otherwise. What stopped in 2023 is the showing.

The leak

Some of what internal counting looks like surfaced this month anyway. BELPOL, the association of former security officers, published internal documents in which the work of the big industrial holdings, MAZ, BelAZ, MTZ, Gomselmash and Amkodor, is rated “unsatisfactory.”

A leak is one data point, unverifiable by design, and we treat it accordingly. Its direction is still worth recording, because it points the same way as everything else in this piece. These are the plants that carry the official industrial narrative, and their public numbers used to be checkable against the fiscal chapters that no longer exist.

Meanwhile, the discourse

Whatever gets removed from the statistical record, the information space keeps producing signals, and they can be measured. Our economic-signals agent processed the H1 media flow on eleven indicators. Three stand out.

Inflation: 2,059 documents, sentiment at -0.89 on a scale from -1 to +1. That is the second half-year running at this depth, and the negativity holds in the state segment as well as the independent one. Price growth appears to be the one economic fact no editorial policy has managed to reframe.

Business climate is the only indicator at critical alert. Coverage grew 111% over the half-year with sentiment at -0.43, and state media are themselves among the top sources, mostly through coverage of inspection campaigns and regulatory pressure on what remains of private business.

Labor migration sits at -0.95, the most negative reading on the board. We wrote about that story separately in The Cure Before the Diagnosis.

Why we’re pairing these layers

The point of putting official statistics and discourse data in one database is not to catch a bad quarter. It is that concealment becomes something you can date and describe. For the fiscal chapters we can now say exactly which tables existed, through which edition, and when they stopped; and the media record shows what the surrounding information space was registering while they were dark. The state stopped publishing its finance pages in 2023. The discourse has priced the economy every week since, and the price it shows is negative.

We picked the economy as the pilot because July happened to supply all three layers at once: a print, a leak, and a blackout. The full accounting of the blackout, every vanished section, every frozen indicator on the state data portal, every retrospective edit across three decades, is a longer research paper coming out on FactCheck.LT.

Method notes. Yearbook findings come from an automated cross-edition diff of 13 editions of “Belarus in Figures” (2014–2026) in the FactCheck.LT Belstat corpus (988 publications, 5.9M extracted observations); section-level disappearances were checked against the source PDFs. Only about 2.4% of the corpus’s figures were ever published in machine-readable ruled tables; the rest required positional text reconstruction, which is why we cite structural findings here rather than fine-grained values. GDP figures are the official H1 2026 releases as reported. BELPOL documents are cited as published by the association, without independent verification. Discourse sentiment: FORESIGHT economic-signals agent, H1 2026 run.

On 22 June, the BIPART research centre published The Presidential Administration within the Belarusian Governance Structure: Functions, Personnel, and Influence — forty-six pages in Russian and English on the one Belarusian institution that combines maximum influence with minimum visibility. Anyone working on Belarusian governance should read it. It is the most complete open-source account of the Administration we have, and parts of it are simply new.

It is also a study whose narrative runs slightly ahead of its own material. Below: what it establishes, where its central claim outruns its evidence, and four things we can measure to settle the difference.

What the study establishes

Three contributions are durable.

The first is a legal genealogy. The study traces the Administration through the decrees that made it — No. 6 and No. 23 of 1994, which created a “working body”; No. 97 of 1997, which reclassified it as a “state administration body” and handed it the personnel register, ideology and the state media; No. 40 of 2017; and No. 303 of 14 August 2025. That last document matters most, because it is classified for official use, and the study reconstructs its content anyway: the Administration now handles the president’s interaction with the All-Belarusian People’s Assembly, presidential assistants may call in the security agencies and initiate disciplinary proceedings against officials, and regional inspectors are formally charged with political stability, ideological quality and oversight of civic and religious organisations. This is the clearest published account of what the 2025 reorganisation did.

The second is prosopography. Eleven heads since 1994, with tenures, prior posts, education and subsequent careers, plus a table of former Administration staff now running ministries and a typology of the recurring career loops — foreign ministry to Administration to foreign ministry, information ministry to Administration to state television. As a dataset for elite network analysis this is immediately reusable.

The third is de-mythologisation, and it is the study’s most useful move. The head of the Administration is not the “second person” in the state: he ranks seventh in the formal register, and the 2022 constitutional amendments route emergency succession to the chair of the Council of the Republic. The All-Belarusian People’s Assembly, described by a sitting deputy head of the Administration in March 2026 as “insurance against unforeseen circumstances”, is a reserve arrangement rather than a power centre. And economic policy still belongs to the government and the National Bank, who “risk losing their posts” for it. Each of these corrects an assumption that circulates freely in exile analysis.

The second ledger

The study reads its material as a single upward curve: a working body in 1994 becomes the pivot of the system by 2025. Every decree adds. But the same forty-six pages contain a second, unaggregated list.

Caption: Every decree the study examines added functions to the Administration. The study also records, in passing, six removals and vacancies it does not add up.

The assistant for national security was abolished in 2021 after Viktar Lukashenka vacated it. The assistant for legislative and judicial affairs — created by the 2017 decree the study reads as an expansion of legal control — was dropped from the structure in 2025. The assistant for the financial and credit system has been vacant since 2023, the assistant for oversight of strategic projects since 2024, and one of four deputy head posts was still empty in May 2026. The 2025 decree also quietly removed the clause making the head of the Administration an ex officio member of the Council of Ministers and its Presidium. Meanwhile the establishment fell from roughly 169 posts to 118 in 2017 and to 114 with civil-servant status by June 2025.

The study’s explanation for the empty economic posts is that “the corresponding functions may have been redistributed” internally, and that under sanctions Lukashenka may prefer non-public administration of these issues. Perhaps. But that is a hypothesis stated in the conditional, and there is a competing one: an institution whose written mandate keeps widening while its staff shrinks and four senior posts stand empty for years may be losing capacity rather than gaining it — narrowing to a political-control core while economic coordination migrates to the government and security coordination to the State Secretariat of the Security Council and the family circle. The study does not test either reading. It does not have to. But it should not present only one of them as description.

Chokepoint or decision centre?

The abstract calls the Administration “the pivotal institution within Lukashenka’s system of power”. The body of the study documents something more specific and more interesting.

Caption: Sorted by the study’s own evidence, the Administration’s grip is a function of what is being controlled.

Where the object of control is officials, the Administration is dominant: it screens, vets, interviews and places, and every appointment on an 850-post register passes through it. Where the object of control is outcomes, it is a filter and a mediator — the study’s own formulation for the economy is that the Administration helps the government and the National Bank “promote their initiatives before Lukashenka” by framing them acceptably. Even in ideology, the study concedes that content is set by Lukashenka’s cues and that the press secretary who shapes his image reports to him directly, not through the Administration. And the closing pages say plainly that many key decisions are taken not in the Administration but in a narrower circle including the family.

That is not a pivotal institution. It is a procedural chokepoint with an unusually good seat. The distinction matters for anyone using this research operationally: sanctions design, accountability mapping and transition planning all depend on whether you are looking at where decisions are made or where they are processed.

Eleven heads, and the trouble with small numbers

Caption: Median tenure is two years and six months. The study’s mean appointment age of 49 rests on eleven cases spanning sixteen years of age.

The biographical chapter is the study’s best data and its weakest inference. Four of eleven heads came from the security services, three from economic agencies; the study also reports that four were born in Vitsebsk region and that the average appointment age is 49. With an N of eleven and no base rates, these are descriptions of a table rather than findings about a system — as the study’s own interviewees say, when they observe that origin and profession are not decisive and that Lukashenka picks “the strongest individual he sees at any given moment”.

One detail the chapter records but does not discuss: Siarhieienka left on 29 March 2024 and Krutoi was appointed on 27 June 2024. For three months, in a system the study describes as running through the Administration, the Administration had no head. That gap is worth a paragraph.

What we would want next

The study’s method statement is one sentence: it uses open sources and expert interviews. We are not asking for the author’s name or the interviewees’ — exile research on the Belarusian nomenklatura has good reasons for both silences. But the analytical weight here is carried by quotations, and readers cannot tell how many people were interviewed, when they last had access, or whether five separate claims come from one former official. Indexed anonymous sources (”Interview 3, former ministry official, 2019–2023”) would cost nothing and change the evidentiary status of the whole document.

Two absences are more substantive. There is no comparative frame — the “administration as parallel government” pattern is well described for Russia and Kazakhstan, and Belarus would look either typical or distinctive against it. And for a study of an institution said to be growing, there is almost no behavioural data: no output volumes, no budget, no measure of how much of the Administration’s normative production is now classified. The growth is documented in the wording of decrees, not in observable action.

Four things we can measure

Our corpus covers state and regional media, the pro-government Telegram layer and official publication channels. Four indicators would discriminate between the two readings above:

1. Vacancy persistence. Whether the deputy head post and the two economic assistant posts are filled, tracked through personnel-day coverage. Sustained emptiness is evidence of narrowing, not redistribution.

2. Attribution ratio in economic coverage. How often state media put economic decisions in the mouths of Administration principals versus government and National Bank principals. Figure 2’s bottom row predicts a stable government-heavy ratio.

3. Transmission lag for the ideological agenda. The Administration’s monthly “Materials for Members of Information and Propaganda Groups” set a single theme — May 2026’s was tourism development. Measuring how many days pass before that theme surfaces in regional state outlets is a direct test of the claim that the Administration sets the ideological agenda rather than relaying it.

4. Regional inspector visibility. Public activity of the seven presidential assistants serving as regional inspectors, against that of the oblast executive committees they shadow. The 2025 decree gave the inspectors a supervisory mandate; visibility shows whether it is being exercised.

Three dated claims

To be checked here on 30 June 2027:

1. At least two of the three senior Administration posts vacant in May 2026 — one deputy head, assistant for the financial and credit system, assistant for oversight of strategic projects — will still be vacant.

2. In state-media coverage of economic policy decisions, government and National Bank principals will continue to outnumber Administration principals by at least three to one.

3. Dzmitry Krutoi will still be head of the Administration, taking his tenure past the eleven-head median.

If all three hold, the Administration is a stable processing centre with a shrinking establishment, not an institution accumulating power. If the first two fail, BIPART’s reading is the right one and ours is wrong.

BIPART’s study is available in Russian and English. This review draws only on the published text.

Last month the Center for New Ideas published Monitoring the Belarusian Transition, a report by Ryhor Astapenia, Dzmitry Kruk and Anton Radniankou built on a survey of 29 experts conducted in April 2026, together with an interactive scenario tree at newideas.center/transitmonitoring. Three headline medians: a 15% probability of a power transition by 2030, 60% by 2035, 90% by 2040. A base scenario: authoritarian succession. And a warning that the departure of the incumbent is a moment of maximum uncertainty rather than an opening.

We think this is the most useful Belarus transition document published in years, and we think its numbers should be treated with more suspicion than they will be. Both of those judgements come from the same place: we have spent the past year measuring how well predictions about Belarus have actually performed.

What the report gets right

Start with the framing, because it is the contribution that will outlast the percentages. The report defines transition operationally — a change of the first person and the transfer of power to a new decision-making centre — and refuses to equate that with democratisation. The comparative base rate is not a caveat at the end; it is the premise. Most personalist autocracies produce another authoritarian regime when the autocrat exits, and the report builds its scenarios on that fact rather than around it.

Second, the methodological housekeeping is better than the genre norm. Medians rather than means, with the reason stated. The panel’s professional composition disclosed. Different bases reported for the open-ended items. Non-responses counted. A separate dataset file. Named reviewers.

Third, the recommendations for the EU are institutionally literate in a way that advocacy documents usually are not. A Belarus Democratic Transition Facility modelled on the Ukraine Facility, as a distinct regulation with objectives, access conditions and a payment mechanism, plus NDICI grants, EIB and EBRD access and a pre-agreed sanctions-relief sequence. That is a proposal a desk officer can move. “The EU should pay more attention” is not.

The measurement problem

Now the suspicion. Fifteen, sixty and ninety are suspiciously round numbers. The report acknowledges in a footnote that variance across the panel is high, but the variance is never shown: no interquartile range, no minimum and maximum, no distribution. A median without a spread converts disagreement into false precision, and it is these three figures, not the dispersion behind them, that will be quoted for the next two years.

The panel’s composition also sits awkwardly against the report’s own conclusions. The behaviour of the security services is named as the single most critical uncertainty across every case examined, and as the second most underestimated factor overall. Two of the 29 respondents declared security and armed forces as their field. Economic dependence on Russia is named as the primary constraint on any successor. Two respondents came from economics and finance. Twenty of 29 are political scientists and international relations specialists. How the 29 were selected is not disclosed at all.

Then there is the tree. A survey that asks marginal questions — which trigger, who takes power, what Russia does — cannot by itself yield the conditional probabilities a 14-branch tree requires: the probability that the siloviki fracture given a health trigger and a weakened Russia. The interactive page does not say whether those conditionals came from the experts, the authors, or the model. It does display “Σ P = 1.0” and scales line thickness to path weight, which lends an interpolation the visual authority of a Bayesian object. The claim that the two axes were validated by factor analysis of expert responses deserves the same scepticism: with n=29 and a mix of categorical and percentage items, factor analysis is at the edge of what the method supports.

To the authors’ credit, the page discloses that the interactive model was built by Claude Opus 4.8 under their direction. We run LLM-based analytical pipelines ourselves, so we will say plainly what that disclosure implies and the page does not: a language model asked to construct a transition tree reproduces the central tendency of the transition literature. That is the same literature that formed the panel. The panel and the model are not independent sources, and the tree reads as their convergence rather than as a check on either.

What our audit says about forecasts like this one

This is not a hypothetical concern, because the historical record on Belarus predictions is measurable and we measured it. Working from two archives of English-language commentary on Belarus, we extracted 715 predictive claims and scored 157 of them against what actually happened. Overall accuracy: 65 out of 100.

The interesting part is the spread by analytical framework. Organisations that simply documented what the regime was doing scored highest: Reporters Without Borders averaged 88.3 across 12 predictions, Belapan 91.0 across 3, Reuters 86.5 across 2. International institutions and mainstream outlets hedged their way to the middle, around 57. Think tanks applying democratic transition models did worst: the Jamestown Foundation averaged 18.3 across 3 predictions, RFE/RL 49.7 across 6. The small samples per source deserve stating — three predictions is three predictions — but the pattern holds at cluster level. Predictions grouped around authoritarian consolidation scored 6 correct and 0 wrong. Predictions grouped around European transition scored 3 and 4.

The distance between an analyst and the facts, and the closeness to transition theory, predicted error better than access or expertise did.

Read the new report against that record and the verdict is split rather than dismissive. Its base scenario — authoritarian succession, power passing to actors already inside the system, cosmetic liberalisation without structural reform — belongs to the cluster that historically scored 100%. Its treatment of the democratisation window as narrow, conditional and improbable is exactly the correction the record demands. Where it becomes speculative is where the record says to be most careful: in assigning calibrated-looking probabilities to trigger events. Our audit also found that 10 of 12 significant post-2013 developments were complete blind spots in the earlier prediction corpus, and that false thaws were identified in 2007, 2009 and 2010. The trigger question is precisely the question the genre has historically got wrong.

The contradiction at the centre

One finding in the report deserves to be pulled out and read twice. Asked which factor analysts most underestimate, the panel’s most frequent answer was the capacity of Belarusian society to act. Asked what society will do at the moment of transition, 55% said it will be passive and accept the status quo — and all four scenarios are built on that passivity.

The single variable the panel identified as underestimated is the one the model holds constant. “Experts underestimate X” is a diagnosis of the panel. It should have been applied to the panel, not printed beside its output.

What monitoring would need

The report promises a series, and the word in its title is monitoring. As published, it contains no operationalised indicator, no threshold, no update rule and no falsification criterion for its base scenario. Nothing states what would move probability mass. Without that, the next wave measures a change in the panel’s mood rather than a change in the country.

This is the gap we can fill, and we would rather propose than complain. A pre-transition period that consists of the quiet redistribution of resources, connections and influence — the report’s own formulation — leaves traces in the information environment, and those traces are countable. Four candidates, all measurable in our corpus of roughly 3 million documents from 73+ sources:

• Successor salience. The share of succession-framed items in state media in which any single figure other than the incumbent is the dominant subject. A closed managed handover requires a name to be normalised before it is announced.

• ABPA function drift. Whether coverage of the All-Belarusian People’s Assembly emphasises constitutional continuity or personnel. The Turkmen and Kazakh precedents both turned on the dual-power body acquiring an occupant.

• Integration vocabulary density. The frequency of Union State and integration-agreement language in state media, which is the leading edge of scenario four rather than a rhetorical constant.

• Siloviki visibility. Whether security-bloc figures appear as policy voices or only as enforcement subjects. A bloc with ambitions acquires a public register first.

None of these predicts a transition. All of them are observable now, monthly, without a panel — and each one can be scored against the scenario tree rather than argued about.

Three of our own, dated

Criticising someone else’s probabilities obliges us to post some. Three predictions, each with a resolution criterion and a review date of January 2028. We will publish the scoring whether or not it flatters us.

1. Through Q4 2027, no single figure other than the incumbent will exceed 15% of succession-framed mentions in the state-media corpus in any quarter. Falsified by a crossing sustained across two consecutive quarters.

2. Through Q4 2027, ABPA coverage will remain institutional rather than personal: no named individual becomes the dominant subject of ABPA coverage for two consecutive months.

3. Over the four quarters to Q2 2027, integration-agreement vocabulary density in state media will not decline. A sustained decline would be the first corpus-visible evidence that the elite diversification the report describes is actually happening.

If the base scenario is right, all three hold and we learn little. If any breaks, the tree needs redrawing — and we will have the date it broke.

Read the report: Monitoring the Belarusian Transition (Center for New Ideas, July 2026). Interactive scenario tree: newideas.center/transitmonitoring.

On 10 July, the designer clothing brand Honar announced it would close by the 19th. The same day, Ragna — the Minsk ethno-shop that had promoted folk artisans for thirteen years — shut its doors on Sukhaya street. Two days later, LSTR Adzieńnie, the streetwear label that had put historical and national prints on Belarusian shoulders since 2011, announced its own closure — days after presenting a new print for Kupalle, the midsummer festival. That festival, meanwhile, was banned outright in the village of Liatokhi near Vitebsk, and “Nash Hrunvald,” the country’s largest historical reenactment event, was cancelled.

None of it, by the available evidence, was voluntary. The Viasna human rights centre reports that the owner of one of the venues was detained by security officers, interrogated in detail about his personal contacts and social media, and released only after agreeing to liquidate the business.

And in the same week, across Vitebsk, the state opened the Slavianski Bazaar festival, where Lukashenko personally presented Union State prizes in literature and art and invoked the brotherhood of the Belarusian and Russian peoples.

The confession

Usually, attributing intent to a purge requires inference. Not this time. On the pro-regime STV channel, host Ryhor Azaryonok — joined by MP Vadzim Hihin — explained the closures on air, in his own words: “All these attempts of yours to sell little t-shirts with half-hints — we see all of it. As we once incinerated the Knihauka bookshop, so we will continue... if you hope to drop anchor somewhere, dig in your teeth somewhere — we will knock your teeth out.”

Strip the menace and a precise doctrine remains: the regime reads apparel as identity signaling. A vyshyvanka print is a “half-hint.” A Kupalle t-shirt is a political act. The security services’ interrogation questions — contacts, social media — confirm the same reading: this was not commercial regulation but the removal of nodes in an identity network.

The July closures extend a chain running back to 2020: Symbal.by (its founder Pavel Belous still imprisoned), the Hrodna shops Admetnasць and Tsudounia, Orsha’s Tsudounaya Krama, Minsk’s Moy Modny Kut (now operating in Warsaw as Vyraj), the Knihauka bookshop Azaryonok boasted of “incinerating.” What is new is the tempo — three brands in ten days — and the explicitness of the accompanying commentary.

What our data adds: the purge follows a measured defeat

Here is where the corpus earns its keep. In our H1 2026 review we reported the most counterintuitive finding of the half-year: in Q2 2026 — the Victory Day quarter, when Soviet-nostalgic framing historically peaks — Soviet nostalgia collapsed from 39.8% to 23.4% of identity discourse, while “European Belarusian” identity surged from 34.2% to 43.1%, retaking a commanding lead across a corpus of ~3 million documents.

Read the July purge against that curve and the timing stops looking incidental. The regime is losing the semantic field of Belarusian identity — measurably, quarter by quarter — and it is responding where it still holds unchallenged power: in the physical layer. Unable to out-narrate “European Belarusianness,” it closes the shops that sell it, cancels the festivals that perform it, and bans the rituals that embody it. Azaryonok’s “half-hints” line is the doctrine’s plain-language version: the discourse cannot be won, so the infrastructure must go.

The weekly topic data adds a coda: in the culture topic, state media out-published independent outlets 42 to 23 this week (+30% state balance). On the surface, the regime dominates cultural coverage. But that dominance is Slavianski Bazaar and Union State prizes — the celebration track — while the identity track it cannot control is handled by the police.

What this does to the balancing thesis

Our H1 report argued that Belarus has entered “managed balancing”: releases continue, dialogue channels stay open, discursive control softens (the civil-society Substitution Index fell to 0.86). The July purge does not contradict that thesis — it refines it, in the instrumental direction we flagged. The regime can afford to loosen its grip on discourse precisely because it is tightening its grip on discourse’s material substrate. Semantic softening for the Western interlocutors; physical dismantling for the domestic identity infrastructure. Both halves of the tightrope, walked simultaneously.

For our Predictions Ledger, this cuts both ways. Prediction #8 holds that “European Belarusian” framing above 40% signals the balancing configuration is intact — and at 43.1% it is. But the purge introduces the mechanism that could bend the curve: if the infrastructure removal succeeds, the identity discourse it sustained may erode with a lag. Whether measured Belarusianness in the information space can survive the demolition of its physical carriers is now, involuntarily, a live experiment — and we are positioned to measure it, quarter by quarter.

Sources: brand announcements via Instagram; CityDog, Zerkalo, Belsat, Reform.news reporting (13–19 July 2026); Viasna human rights centre on the circumstances of the closures; STV broadcast statements as reported by CityDog. Identity-frame data: FORESIGHT corpus (~3,5M documents, 100+ sources), national identity agent, Q1–Q2 2026 runs. Belarus Brief / FactCheck.LT, Vilnius.

On 14 July, the three EU borders with Belarus told three different stories. Latvia recorded 137 attempted illegal crossings in a single day. Lithuania recorded one. Poland recorded zero.

This is not a fluctuation. It is the visible edge of a full re-targeting of the instrument Minsk has operated since 2021 — and the official border statistics of all three countries, read together, date the pivot almost to the week.

The displacement, in official numbers

Poland’s interior ministry reports roughly 235 attempted crossings from Belarus in the first half of 2026 — against more than 15,500 in the same period of 2025. A drop of over 98%, which Warsaw attributes to its border fortification and a claimed 100% apprehension rate. Lithuania’s count for the year stands near 900. Latvia’s, as of 9 July: 7,634 — already 800 more than in the same period last year, on a 173-kilometre border, the shortest of the three.

Whatever share of Poland’s decline is deterrence, the flow did not disappear. It moved. Latvia’s border guard chief Guntis Pujāts stated it plainly: Latvia is “currently the main destination” of the route.

The ramp: from a frozen January to a record July

The Latvian monthly series shows the pivot’s timing with unusual precision. January and February 2026: 12 attempts combined — a hard winter shut the route almost entirely. Late March: the streak begins — since then, Latvia has not logged a single day without an attempted crossing. April: 1,700. May: 2,433. June: 2,263. On 1 July, a single day brought 232 attempts — a figure unseen on any Belarus-EU border since September 2023.

The seasonality is real — warm weather always raises the numbers. But seasonality does not explain the distribution: the same summer that lifted Latvia’s counts to records left Lithuania’s near-flat and Poland’s at zero. The routing is a choice, and Latvian authorities describe the mechanics of that choice directly: migrants arriving legally in Belarus, transported to the frontier, escorted by Belarusian security forces, equipped to breach barriers — and in some cases prevented from turning back.

Why Latvia

Three explanations circulate, and they are not mutually exclusive. Latvia’s border guard chief has named the country’s support for Ukraine among the reasons it became the primary target. The infrastructure gap is real: Latvia’s fence and surveillance build-out lags Poland’s completed fortification, making its 173 kilometres the path of least resistance. And the calendar offers a third: Latvia holds parliamentary elections on 3 October — and the 2021 playbook showed how effectively engineered migration pressure converts into domestic political stress in a campaign season.

The week the response escalated

The past week marked a qualitative shift in Riga’s counter-moves, tracked across our corpus:

The Prosecutor General’s office opened a criminal case against officials of the Belarusian border service, qualifying their conduct since summer 2021 as the “deliberate and systematic unlawful activity of a criminal organization” — the first attempt by a target state to attach individual criminal liability to the hybrid operation’s operators. The interior minister called the situation a “Minsk hybrid attack” and requested 100 border guards from neighboring states. And the pressure is already producing secondary effects: Lithuania is weighing temporary controls on its Latvian border against secondary movement — while the Lithuanian and Latvian presidents jointly rejected internal Schengen controls as “capitulation,” insisting the answer lies at the external frontier.

That last exchange is the strategic point. The instrument’s purpose was never merely to push people across a line; it is to force EU states into visibly un-European measures against each other. A criminal case against Belarusian border officials targets the operators. Internal Schengen checks would hand Minsk the outcome.

What to watch

A checkable expectation for the coming months: if the targeting logic is electoral, Latvian border pressure should hold or intensify through the October 3 campaign and slacken after — the way pressure on Poland tracked its own political calendar in earlier phases. If instead the driver is infrastructure, the completion of Latvia’s barrier build-out through 2026 should bend the curve regardless of the election. Two hypotheses, one border, a measurable difference. We will track both in Weekly Signals.

Sources and method: official figures from the Latvian State Border Guard and Interior Ministry (via LSM), the Polish Interior Ministry, and Latvian border guard chief G. Pujāts’s public statements (9 July 2026). Monthly values for January–March 2026 are approximated from cumulative official reports. Discourse context from the FORESIGHT corpus (~3,5M documents, 100+ sources), which tracked the escalation week of 13–19 July across Latvian, Lithuanian, Belarusian state, and Belarusian independent coverage. Belarus Brief / FactCheck.LT, Vilnius.

Most analysis of EU–Belarus relations reads what Brussels decides. We read what Brussels says — all of it. The EU Neighbours East portal (euneighbourseast.eu) is the European Union’s official communication channel for the Eastern Partnership region: news, statements, sanctions announcements, grant calls, human stories, a youth ambassador programme. We scraped its Belarus section completely — every listing page, every item, 2016 through July 2026 — and loaded all 2,129 documents into the FORESIGHT corpus.

One channel, ten years, zero gaps. Here is what the dataset shows.

1. The attention arc: the real response came a year late — and has since compressed to a third

The 2020 protests doubled the portal’s Belarus output (121 items against 72 in 2019). But the institutional surge arrived only in 2021: 513 items, driven by the sanctions waves, the Ryanair interception, and the border migration crisis. Bureaucracies communicate on bureaucratic time.

Then the pattern most Belarus watchers will recognize: Russia’s full-scale invasion of Ukraine in 2022 did not crowd Belarus out of this channel (494 items, nearly flat). The compression came later and steadily — 319 in 2023, 209 in 2024, 229 in 2025. The portal now runs at roughly a third of its peak, though still well above the pre-2020 baseline of 60–70 items a year.

A necessary caveat before anyone quotes this as “the EU forgot Belarus”: this is one communication channel, not EU policy. EEAS statements, Parliament resolutions, and Council decisions travel through other pipes. What this curve measures is the public-facing communication effort aimed at the region itself — and by that measure, Belarus commands a third of the attention it did in 2021.

2. Belarus is a political story here, not a grants rubric

The portal’s front pages give the impression of a funding notice board. The full corpus says otherwise: pure grant-and-opportunity items are only 17% of the archive (370 of 2,129). The core is news (45%) and analytical publications (20%). On the EU’s own regional channel, Belarus figures primarily as a political subject — sanctions, political prisoners, statements of concern — with the funding layer wrapped around it, not the other way about.

3. The language map: who each genre is for

The portal publishes in English and translates selectively into Russian. That selectivity, mapped across 2,129 items, amounts to an audience declaration:

Everything aimed into the region gets translated: explainers (89% have an official Russian version), human-interest stories (87%), opinion polls (80%), news (78%), grant calls (70%). Two genres conspicuously do not. Analytical publications stay overwhelmingly English (17%) — the reports are a genre written for Brussels and the expert community, not for the societies they describe. And the Young European Ambassadors blog is English-only almost without exception (1 of 271 items): the EU’s youth outreach programme for the region operates, linguistically, as an English-language club.

There is a temporal layer too, visible in the first chart: the discipline of Russian translation has slid from 73% of items in 2021 to 35% in 2026. Two readings fit the data, and we can’t separate them from this corpus alone: a resource shift as the portal’s Russian editorial capacity lost priority, or a deliberate de-russification of EU communication. Both would produce the same curve; naming only one would be storytelling, not analysis.

Why this corpus matters beyond the three findings

FactCheck.LT has published monthly monitoring of how Belarusian state media talk about the EU and its neighbours since 2025. This dataset supplies the missing mirror: how the EU talks to Belarus, over the full decade, in a form that can be queried, compared, and joined against the rest of the 3-million-document FORESIGHT corpus. When Minsk’s media claim that “Brussels has written Belarus off” — or, in the other mode, that “the West is obsessed with Belarus” — there is now a measurable baseline for what the EU’s actual communication effort looks like.

Methodology: complete crawl of euneighbourseast.eu/country/belarus (143 listing pages, 2,129 items, no gaps), July 2026. Genre labels are the portal’s own taxonomy; the news category was introduced in 2021, so cross-genre comparisons spanning that year carry a caveat, while total-volume comparisons do not. Translation shares for 2016–2020 are omitted from the chart due to small samples. 2026 is a partial year (through July). Publication dates are the portal’s published-time metadata (±1 day vs. displayed dates). Part of the FORESIGHT corpus (~3M documents, 73+ sources) built by FactCheck.LT, Vilnius.

Here is a number I cannot stop thinking about. We just finished coding every article Britain’s biggest tabloid, The Sun, has published about Belarus in ten years: 123 pieces, 2016 to 2026. In eight of them, Belarusian society acts, speaks and has motives of its own. Eight articles in a decade.

Not eight sympathetic articles. Sympathy is abundant: repression is actually the single largest frame in the corpus, 25 articles about crackdowns, political prisoners and executions. But sympathy is not agency. In the tabloid’s telling, Belarusians overwhelmingly appear as objects: people things are done to, on a territory things are done through. Across the 83 articles genuinely about Belarus, the country is an instrument of someone else’s will in 54 and scenery in 18. The “someone else” needs no introduction.

The exception is what makes this a story rather than a complaint. For roughly two months, August and September 2020, the frame broke. Maryia Kalesnikava tore up her passport at the border, and The Sun wrote about a woman who out-manoeuvred her own kidnappers. Crowds filled Minsk, and The Sun wrote about a society, not a regime. Even its Lukashenko coverage of those weeks portrayed a man scrambling to react to Belarusians, rather than Belarusians existing at his pleasure. Four of the nine articles from that window grant agency. That is more than the following five years combined.

Then came 2021, the year Belarus finally got the tabloid’s full attention: 39 articles, a third of the entire decade, driven by the Ryanair hijacking, the weaponised migration crisis and the Olympics. And here is the twist the data delivers: the peak of attention was the peak of voicelessness. Of the 34 substantive articles that year, exactly two grant Belarusians agency, both about sprinter Krystsina Tsimanouskaya, the one 2021 protagonist who acted on camera, refusing to board a plane home. In 32 of 34 articles, Belarus is simply identified with Lukashenko. Raman Pratasevich appears in six articles and acts in none. Attention, it turns out, is not the same as visibility.

The pattern has held ever since. After February 2022, Belarus dissolved into war coverage as a launchpad and a nuclear parking lot. In 2025, an election year, The Sun published exactly one Belarus article, and it was about a prisoner release framed as a favour to Trump. The presidential election itself, a story featuring a dictator voting with a Pomeranian named Umka, did not clear the bar. We checked three independent ways; the silence is real.

What did the eight exceptions have in common? Not politics. The earliest is from 2016: a Minsk software engineer who built his father a prosthetic arm in a basement and posted the blueprints online for free. Then a Eurovision duo. Then Kalesnikava. Then Tsimanouskaya, twice. Then Ales Pushkin, the artist whose protest was his art, and whose death in prison The Sun covered as the story of a man who acted. Then Sviatlana Tsikhanouskaya, twice, most recently last week.

Every one of them is a named person with a deed and a legible motive. That is the entire formula. Elections do not clear the tabloid threshold; statements do not; anniversaries do not. A person who does something does.

For everyone working to keep Belarus visible in the West, that is both a constraint and an instruction. The mass-market frame will not carry abstractions, and it will flatten the country into one man’s name whenever we let it. But it demonstrably makes room for a Belarusian who acts. The window of 2020 was not opened by editors. It was opened by people.

The full study, with the corpus, codebook, charts and methodology, is out now on FactCheck.LT: A Country Without a Voice: Ten Years of Belarus in The Sun.

Numbers in this piece: 123 articles (2016-2026), 83 substantively about Belarus; agency in 11, of which 8 belong to society; 2021: 39 articles, agency in 2, Belarus-equals-Lukashenko in 32 of 34; 2025: 1 article. Methodology, reliability checks and editorial adjudications are documented in the full publication.

For a state that spent years mocking Europe’s migration policies, this was a remarkable announcement. But it was not a sudden one. We traced the preparation through 2.96 million documents — and found a six-month normalization campaign that laid every piece of infrastructure for this moment, while never admitting the problem it was built to solve.

Two Parallel Coverages

Between January and June 2026, our corpus contains 253 independent-media documents on labor shortage and workforce topics, led by Charter97 (53), Zerkalo (48), and Euroradio. Their story: exodus. Emigration of specialists, personnel deficits, workers leaving for Poland and Lithuania.

Over the same period, state media produced 158 documents on the same keyword field. But read them, and they tell a different story entirely — not one of departure, but of arrival. Happy foreign workers. Rising migrant statistics. School enrollment guides. In six months of state coverage, the causal chain — workers are leaving, therefore foreign workers are needed — appears exactly zero times.

The diagnosis was taboo. The cure was advertised daily.

The Normalization Timeline

The preparation is visible week by week:

January 8 — BELTA reports amendments to the citizenship law under review at the Council of the Republic’s expert council: simplified paths for “certain categories of foreigners.” The legal infrastructure moves first.

January 22 — BELTA publishes “Happy to work and earn”: a feature about African workers on a farm in the Lida district. The normalization imagery begins.

February 5 — BELTA: “About 3,500 labor migrants arrived in Gomel region in 2025.” Arrival becomes a positive statistic.

March 2 — The trade-union outlet Prof1 runs “Germans retire, and foreigners take their place” — the European precedent frame. What Belarus mocked in 2021, it now cites as normality.

March 9 — Lukashenko receives the Uzbek ambassador. The CTV headline: “Belarus is always glad to see hardworking people from Uzbekistan.” State coverage of Uzbekistan spikes to 130 documents that week — the Mirziyoyev visit enters preparation four months before it happens.

April 1 — BELTA: “Labor migrants from 30 countries work in the agro-industrial complex of Grodno region.” Scale is normalized.

June 5 — The regional outlet Mlyn publishes a practical guide: “Children of foreigners in Belarus: how to enroll in school.” A month before the “with families” invitation, the family-migration infrastructure narrative is already in place.

July 7-8 — State coverage of the Uzbek-labor intersection jumps from 2 documents a day to 14, then 27 — the information ramp precedes the talks by 48 hours.

July 9 — The invitation.

Weekly state-media Uzbekistan coverage with normalization milestones. The March 9 ambassador meeting and the July 9 visit bracket the campaign.

The Ones Who Called It

Independent media did not just document the exodus — they spotted the pivot. On January 11, Nasha Niva published a piece whose title needs no improvement: “Import of People: Belarus spent years criticizing Europe over migrants — now it moves down the same path.”

That was published three days after the citizenship-law story and six months before the presidential invitation. The analytical layer of the Belarusian independent media ecosystem saw the trajectory when it was one data point old.

Our own FORESIGHT system flagged the same track from a different angle: the European Choice agent’s Q1 2026 assessment listed an active “Uzbekistan track” among the regime’s diversification moves, months before the visit was announced.

What “With Families” Means

The invitation’s most consequential phrase is its quietest. Seasonal labor migration does not come “with families.” Family migration is permanent migration — a demographic policy, not a staffing fix.

Belarus has lost hundreds of thousands of residents since 2020 — to repression-driven exile, to war-adjacent emigration, to the slow drain of specialists. A regime that cannot invite its own citizens back is inviting replacements. The school-enrollment guides were not a coincidence; they were the operational layer of a decision that had already been made.

The Coming Pivot — a Testable Prediction

Our economic signals agent rated labor migration the most negatively framed indicator of H1 2026 (sentiment -0.95) — driven by independent coverage of the exodus. Here is our prediction, stated now so it can be scored later:

By Q4 2026, the labor migration indicator’s sentiment will turn positive — not because the exodus ends, but because state media will flood the field with arrival stories: integration successes, grateful workers, growing statistics. The state will sell the solution to a problem it never acknowledged having. When that happens, the sentiment flip itself will be the measurement of the propaganda campaign’s scale.

We will check in January 2027.

Methodology: FORESIGHT processes 3 million documents from 80+ sources across the Belarusian information space. This brief draws on keyword-tracked coverage of labor and migration topics (January 1 – July 10, 2026), weekly Uzbekistan mention counts segmented by source type, and manual review of all 158 state-media documents in the sample. FactCheck.LT, Vilnius.

Three signals

1. In-country civil society goes quiet — Civil society (in-country) recorded a 29% week-on-week decline against its baseline, 24 percentage points below the corpus drift. No other domestic actor fell as steeply. One possible reading: the Independence Day security environment — Interior Minister Kubrakov’s pre-holiday security briefing and heightened public-order posture were widely reported — may have suppressed both the activity and the coverage of any organized civic presence inside the country.

2. Pro-regime structures lose ground — Pro-regime structures (GONGO) fell 16%, putting them 11 percentage points below the corpus baseline. For the week of Independence Day, when state-adjacent mobilization would be expected to peak, that relative absence is notable. The dominant regime narrative was built around Lukashenko’s Asian diplomatic tour and the holiday itself; GONGO actors appear to have been crowded out of the frame rather than actively generating coverage.

3. Political opposition (exile) slides further — Political opposition (exile) dropped 19%, a 14-point gap below the overall corpus fall. This is the exile political tier’s weakest relative showing in the current dataset. The Coordination Council did hold an internal event — electing a second deputy speaker for its new convocation — but that procedural milestone generated little measurable coverage volume, suggesting the development did not break into broader information flows.

One to watch

Watch Pro-regime structures (GONGO) in the coming week: Independence Day programming has now concluded, and if GONGO coverage does not rebound toward baseline, it would suggest the dip reflects something more structural than a one-week scheduling effect — a checkable expectation by next Friday’s numbers.

Belarus Brief tracks 12 actors across 3 tiers (states, domestic, exile) weekly. Methodology: corpus-level mention detection with 4-week baseline deltas, ~3 million documents from 73+ sources. Built by FactCheck.LT (Vilnius).

The headline: Belarus has moved from oscillation to managed balancing. The scenario engine’s probability for “managed balancing” jumped from 35% to 48%, while “intensified repression” collapsed from 26% to 13%. Every fragility indicator improved. The civil society discourse softened. Media freedom metrics turned positive for the first time in our tracking. And yet — FIMI volumes hit an all-time record, and the Oreshnik missile discourse doubled.

This is not liberalization. It is a regime learning to walk a tightrope, and getting better at it.

1. Scoring Our April Forecasts: What We Got Right and Wrong

Transparent forecast verification is core to the FORESIGHT methodology. Here is the honest scorecard:

Transparent forecast verification: two confirmed, one not realized, two wrong.

Our biggest miss deserves explanation. In April, we predicted the “democratic institutions” discourse index would fall back as the Q4 2025 anomaly left the analytical window. Instead it rose to 97.1%, and the media freedom dimension flipped from -34.7 to +20.2. We underestimated the durability of the dialogue track: prisoner releases continued, negotiation channels with Washington and Warsaw stayed open, and the regime’s discourse followed. The error was treating Q4 2025 as an anomaly when it was the beginning of a trend.

2. The Balancing Shift: Scenario Probabilities

The scenario engine, fed by fragility, FIMI, and European choice data, produced a dramatically different distribution than three months ago:

Probability mass moved from repression (-13 pp) to managed balancing (+13 pp).

The probability mass moved decisively from repression to balancing. Liberalization, while still marginal, more than doubled — the first time this scenario has moved at all since we began tracking.

3. FIMI: The Convergence Completes

The “fake” vs “disinformation” distinction — populist versus institutional labeling of information threats — is nearly gone. The ratio reached 0.96 in Q2 2026, up from 0.72 in early 2025. Combined volume hit a record 19,256 mentions in a single quarter.

Left: record volumes. Right: the ratio approaches parity (0.96) — the two-term system collapses into one.

What does convergence mean? In 2025, “fake” was the regime’s weapon against internal opponents, while “disinformation” framed external threats. By mid-2026, both terms point overwhelmingly West. The information-defense apparatus has unified its vocabulary into a single external-threat narrative — even as the diplomatic track with the West deepens. This is the tightrope in miniature: negotiate with Washington while narratively mobilizing against “Western disinformation.”

4. Fragility: The Regime Feels Safer

All five dimensions of the Fragility Pentagon improved between the Q1 and H1 measurements:

The teal contour (H1) sits outside the gray one (Q1) on four of five dimensions.

Higher gap ratios mean less compensatory state-media effort: the regime is spending less narrative energy defending its weak points. The largest improvements came exactly where April showed the deepest vulnerabilities — repression reliance and personalization. A regime that releases prisoners needs to talk about repression less.

5. Civil Society: The Capture Softens

The Substitution Index — measuring regime appropriation of the term “civil society” — fell to 0.86 in 2026, the second-lowest value since tracking began in 2021 and well off the total-capture peak of 1.00 (H1 2023, the Bialiatski sentencing period).

The index (dots) retreats from total capture while the neutral share of independent coverage (teal) falls — both sides intensify the contest.

Two readings are possible. The optimistic one: the dialogue track requires loosening discursive control, and independent voices are regaining semantic ground. The instrumental one — suggested by our LLM interpretation layer — is that the retreat is itself part of the operation: a regime demonstrating “openness” to Western interlocutors by softening its most visible discursive monopolies.

A supporting signal favors the second reading: the neutral share of independent media coverage fell from 57% to 44% over the same period. Independent outlets are not relaxing — they are contesting the term more actively than ever. The softening is asymmetric: the regime loosens, the opposition escalates.

6. Economy: Normalization With One New Alarm

Of the eleven tracked economic indicators, the two critical alerts from Q1 — inflation and labor migration — both de-escalated to warning level. Sanctions pressure coverage went quiet (normal). The IT sector continues its anomalous positive-sentiment rise.

The one new critical alert: Business Climate (rising volume, sentiment -0.43). As macro-panic subsides, discourse is shifting to structural complaints — regulatory pressure, inspection campaigns, the operating environment for what remains of private business. This is a slower-burning signal than inflation, but historically a more reliable predictor of elite-level friction.

7. The Media Ecosystem at Half-Year Scale

The 11-source independent media analysis now covers 62,879 publications across H1, clustered into 44,143 events:

• Multi-source coverage rose to 12.2% (from 9.1% in the Q1-only run) — the shared news agenda is growing

• State dependency stable at 4.3% (v3 context-aware detection)

• A structural novelty: STV became the #2 most-cited state source (227 citations), displacing SB — worth monitoring as a possible shift in which state outlets independent media treat as newsworthy

• Belsat remains the volume leader (99% exclusivity); Mediazona remains the most overlapped specialist (47%)

8. The European Choice: The Tightrope, Quantified

With Q2 2026 data complete, the European choice picture resolves — and it complicates any simple liberalization reading:

Eastern turn holds 42–54% throughout. The Q1 2026 multi-vector moment (LLM layer) reversed by Q2.

The Q1 2026 “multi-vectorism” moment — when the LLM layer flagged a qualitative shift and the regime debuted its “transit corridor between the EU and the East” formula — did not consolidate. By Q2, the dominant frame returned to Eastern turn both quantitatively (52.1%) and qualitatively. Our own agent’s Scenario A indicator (multi-vector rising above 15%) failed: it fell to 10.9%.

The Q2 assessment is blunt: partnerships with Russia, China, North Korea, Uzbekistan, the UAE dominate; “European path” mentions are rhetorically defensive (“Belarus is already Europe”), aimed at deflecting Western narratives rather than signaling integration; contacts with Lithuania and Poland are confined to border management.

How does this square with the balancing shift in every other indicator? This is the central finding of H1: the regime has separated its rhetorical vector from its operational track. The words go East — Union State, EAEU, Global South. The actions balance — releases continue, negotiation channels stay open, fragility compensation drops. Eastern rhetoric is not the opposite of balancing; it is its domestic cover. A regime walking a tightrope toward Western sanctions relief cannot afford to say so in its own media.

8a. The Identity Divergence: Europe Wins Where It Wasn’t Expected

The national identity agent delivered the most counterintuitive result of the half-year:

Soviet nostalgia collapsed 16 pp in the Victory Day quarter; European Belarusian identity took a commanding lead.

In Q1 2026, Soviet nostalgia led the identity discourse for the first time in our tracking. In Q2 — the Victory Day quarter, when nostalgia frames historically peak — it collapsed by 16 percentage points, and “European Belarusian” identity retook a commanding lead.

Put both agents side by side and the divergence is stark: the foreign-policy vector went East while the identity discourse went European. The regime appears to be decoupling geopolitical alignment from civilizational identity — “we are European, and we choose the East” — a formula that keeps both the domestic pro-European majority and the Russian ally manageable. Whether this is a deliberate construction or an emergent property of balancing pressures, it is the discursive signature of the tightrope.

9. Forecast for H2 2026

Based on the full H1 dataset, the scenario engine and our analytical layer project:

Base case (48%): Managed balancing consolidates — in actions, not words. The dialogue track survives and continues through possible Ukraine negotiation windows in Q3. Expect further selective releases and continued FIMI escalation against “Western disinformation” — while Eastern-turn rhetoric holds above 45% of vector discourse. Do not expect the rhetoric to soften: H1 demonstrated that Eastern framing is the domestic cover for Western-facing balancing, not its contradiction. The indicator to watch is not the multi-vector share (which failed as a signal in Q2) but the identity layer: if “European Belarusian” holds above 40% while releases continue, the tightrope holds.

Key risks to the base case:

• Oreshnik discourse doubled in H1 (32 → 64 critical-event documents). If deployment moves from discourse to demonstrated capability, the balancing act becomes untenable for Western interlocutors.

• Business Climate critical alert: structural economic friction could force a choice between further economic opening and elite protection.

• NATO-Baltics escalation remains the largest critical-event cluster (134 documents). A border incident is the scenario-breaker neither side controls fully.

What would falsify the base case: repression scenario indicators (mass arrests resuming, SI returning above 0.95, fragility gap ratios falling), or conversely, genuine institutional steps (legal re-registration of liquidated NGOs, media law changes) that would mark actual liberalization rather than managed balancing.

We will score these predictions in January 2027.

Methodology

FORESIGHT is a Multi-Agent System built on PostgreSQL with pgvector, now processing 2.96 million documents from 73+ sources — the corpus more than doubled since April through the addition of new independent outlets, YouTube state-TV transcripts, and expanded Telegram coverage. The H1 2026 full pipeline ran 16 agents in 3.75 hours. Historical metrics have been recalculated on the expanded corpus; where April and July values differ (e.g., European choice frame shares), the current corpus is authoritative and comparisons use recomputed baselines.

In addition to quarterly and semi-annual deep-dives, FORESIGHT produces automated weekly 12-actor monitoring reports via an n8n workflow.