Freedom House Transnational Repression: Intelligence Source Guide

Freedom House maintains the reference case database of transnational repression: incidents in which a state reached across a border to assassinate, assault, abduct, detain or unlawfully deport someone who left it. It counts only direct physical acts, which makes it rigorous and makes it a floor.

At a glance

Source	Freedom House Transnational Repression
Category	Conflict, Crime & Human Security › Transnational Repression & Digital Rights
Homepage	https://freedomhouse.org/report/transnational-repression
Format	`HTML`
Access	Open — no account required
Disciplines	Government Intelligence, Open Source Intelligence
Mission domains	Transnational Repression

Cross-border repression case database. — as catalogued in the platform’s own source registry.

Freedom House began systematically documenting transnational repression in 2021 and has updated the work annually since. The core artefact is a case database of direct, physical incidents of transnational repression committed since 2014: assassinations and attempted assassinations, physical assaults, abductions and renditions, unlawful detentions carried out at an origin state's request, and unlawful deportations and expulsions of people who would face persecution on return. Each case records the origin state that conducted or directed the act, the host country in which it occurred, the year, the category of act, and the targeted person or group, with a narrative account and sourcing. Around the database sits a body of analysis covering the tactics that are not counted: coercion by proxy, in which relatives still inside the origin state are detained or threatened to control someone abroad; mobility controls, including passport cancellation and the abuse of international policing channels to have exiles arrested; digital transnational repression, including spyware and coordinated online harassment; and legal instruments such as in absentia convictions and abusive extradition requests. The material is published as reports and web pages with a downloadable case list, from the same organisation that produces Freedom in the World and Freedom on the Net.

Human rights reporting is usually organised by country: what a state does inside its own borders. Transnational repression breaks that frame, because the act occurs in one jurisdiction, is directed from another, and targets a person who is often a citizen or resident of neither in any settled sense. No other dataset is organised around that triangle. What Freedom House supplies is a consistent origin-state-by-host-country accounting that makes two things visible for the first time. The first is which governments treat exile as an operational problem to be solved abroad rather than a domestic matter to be tolerated — a genuinely diagnostic fact about a regime's threat perception and its willingness to violate other states' sovereignty. The second is which host countries are permissive: where renditions succeed, where deportations happen in defiance of non-refoulement obligations, where police cooperate with an origin state's request without scrutiny. That second axis is the one policy actually turns on, and it is invisible in country-by-country human rights reporting. For an intelligence or law enforcement analyst, the database is also the open-source baseline for a category of foreign state activity that increasingly falls under national security and foreign interference legislation.

Who publishes it, and why that matters

Freedom House is a Washington-based non-profit founded in 1941. Historically the large majority of its funding has come from United States government sources — democracy and human rights programme grants rather than commissioned intelligence work — with the remainder from private foundations. That matters in three ways you should reason about rather than dismiss. First, perception: origin states routinely characterise the organisation as an instrument of US foreign policy, which means findings will be contested on those grounds regardless of their quality, and a product resting solely on this source will be attacked on provenance rather than substance. Second, framing: an organisation funded to promote democracy defines its subject in terms of political freedom, and its coverage of states aligned with its funder is thinner than of states that are not — a bias that is structural rather than dishonest. Third, continuity: heavy dependence on a single government's foreign assistance is a real institutional risk, and reported disruption to United States democracy assistance funding in 2025 affected Freedom House's staffing and programming. Do not assume the annual update cycle is guaranteed. Check the current status of the project on the site before designing a workflow around it.

Provenance is the first question to ask of any dataset and the one most often skipped. Who collects it, what their incentive is, whether they publish a methodology, and whether they correct the record when they get something wrong all bear directly on how much weight a finding drawn from it can carry.

What a record actually contains

The fields you will be working with, what each one means, and whether it is something you can pivot on. Read the meanings carefully — more analysis is wrecked by misreading a field than by failing to find one, and a field that looks like an observation is often an inference.

Field	Type	What it means	Pivot value
`case_id`	string	Identifier for an incident of direct physical transnational repression. The unit is the incident, which may involve more than one targeted individual and does not decompose to one row per person.	Joining narrative accounts to the tabulated record, and tracking whether a case is revised in a later release.
`year`	int	Year in which the incident occurred, with the series beginning in 2014. Cases are added retrospectively when documentation emerges, so a given year's count grows after the fact.	Temporal analysis, but only if you re-download rather than relying on a prior year's file.
`origin_state`	string	The government that conducted or directed the act against its own nationals or former residents abroad. This is the accountability anchor of the whole dataset.	Country dashboards, sanctions and designation research, diplomatic and intelligence service structure.
`host_country`	string	The country in which the incident occurred. Reveals which states are permissive environments, whether through complicity, capability gaps or unwillingness to intervene.	Host-state legal framework, extradition and deportation practice, bilateral security cooperation agreements.
`incident_type`	enum	The category of physical act: assassination or attempt, assault, abduction or rendition, unlawful detention at an origin state's request, or unlawful deportation. Categories are mutually exclusive per case.	Severity and escalation analysis; comparing an origin state's methods across host countries.
`target_group`	string	The community or category targeted — a national or ethnic diaspora, a political movement, a religious minority, a professional group such as journalists. Individual names appear in the narrative where already public.	Diaspora community risk assessment and protective outreach, handled with extreme care about identification.
`narrative`	string	The account of what happened, with sourcing. This is where the evidentiary weight actually lives; the tabulated fields are an index to it.	Primary reporting, court records, government statements, and the journalism the case was documented from.
`sources`	array	Reporting and documentation relied on, typically media, court filings, NGO documentation and government statements. Cases resting on a single source sit at a different evidentiary level from those with several.	Independent verification; retrieving the underlying record before relying on a case in a legal or policy product.
`campaign_designation`	string	Where the analysis identifies a sustained campaign by an origin state against a particular community, as distinct from an isolated incident. A campaign judgement is analytical, not merely arithmetic.	Pattern-of-conduct arguments for litigation, sanctions and foreign interference proceedings.
`non_physical_tactics`	array	Tactics documented in the analysis but excluded from the case count: coercion by proxy, mobility controls, digital harassment, spyware, abusive legal process. Present in the reports, not in the tabulated database.	Spyware forensics research, Interpol abuse casework, family-detention documentation by other organisations.
`interpol_nexus`	string	Whether the case involved international policing channels — a red notice, diffusion or bilateral request — as the mechanism of detention. Recorded in narrative rather than as a structured flag.	Interpol case files and challenge processes, national extradition records, legal defence organisations.
`outcome`	string	What happened to the targeted person: killed, injured, returned to the origin state, released, granted protection. The outcome distinguishes a completed rendition from a failed attempt, which matters legally and analytically.	Asylum and protection case law, prisoner documentation in the origin state, subsequent trial reporting.

Coverage — and what is not in it

Global, from 2014 to the most recent annual update, with several hundred documented direct physical incidents attributed to dozens of origin states across a large number of host countries — figures that grow with each release both because new incidents occur and because older ones are documented late. Coverage is deepest where a free press and organised diaspora communities exist to document what happened, which means Europe, North America, Turkey and parts of Asia are well represented as host countries, while incidents in states with no independent media are systematically under-recorded. Origin-state coverage concentrates on regimes with large exile populations and the reach to act against them. The database counts only direct physical acts, and Freedom House is explicit that this is a deliberate methodological floor rather than an estimate of the phenomenon: the far larger volume of coercion by proxy, digital harassment, passport denial and abusive legal process is analysed but not counted, because it cannot be counted consistently. Update rhythm is annual, with the report and case list published together, and earlier years revised upward as documentation emerges. Treat the current release as authoritative over any archived copy, and treat any year's count as provisional for several years afterwards.

Known blind spots

Absence of evidence here is not evidence of absence. These are the conditions under which Freedom House Transnational Repression will not show you something that is nevertheless real:

Only direct physical acts are counted. The dominant tactics of transnational repression — threatening relatives at home, cancelling passports, filing abusive extradition requests, coordinated online harassment — are documented in the analysis and absent from the database, so the case count is a floor by construction and not an estimate of scale.
Incidents in countries without independent media or organised diaspora networks are systematically missing, which means the host-country distribution partly measures documentation capacity rather than where repression occurs.
Attribution to an origin state requires evidence linking the act to that government. Deniable operations using criminal proxies, local intermediaries or unaffiliated sympathisers are exactly the cases least likely to be attributable and therefore least likely to be counted.
Victims who never report are invisible, and there are strong reasons not to report: fear of retaliation against family, immigration precarity in the host country, distrust of host police, and the reasonable belief that reporting will change nothing.
The unit of counting is an incident, not a person, so a single operation against several people counts once and comparisons between origin states are sensitive to how operations are structured rather than to how many people were harmed.
Recent years are always under-counted at the time of publication because documentation lags events by months to years; a downturn in the most recent year is usually an artefact and is routinely misreported as a decline.
Coverage of states closely aligned with the publisher's principal funder is thinner, which is a structural feature of a US-funded democracy organisation rather than an accusation, and it means the dataset should not be used as the sole basis for comparative claims about Western states.
The database records repression, not the protective response. Whether the host country investigated, prosecuted, expelled diplomats or did nothing is largely absent, which is precisely the variable a policy analyst most needs.
Deportations that are lawful on their face but result in persecution sit at the boundary of the definition, and different reasonable coders would classify a meaningful share of removal cases differently.

Write the blind spot into the product. A statement that something “was not observed in Freedom House Transnational Repression” is defensible; a statement that it “did not happen” is not, and the difference is what survives cross-examination.

Access, licensing and what you may do with it

Access model: Open — no account required

The reports and the accompanying case data are published openly on the Freedom House website without registration or payment, alongside the organisation's other country assessment products. The case list has generally been available as a downloadable file or a browsable web view published with each annual report; formats and hosting have varied between releases, so retrieve what is actually there rather than assuming continuity with last year's structure. There is no API. The correct workflow is an annual manual retrieval, archived with the release identifier and retrieval date, plus reading the narrative report, which contains the methodology, the definitional boundaries and the analysis of uncounted tactics. Do not build a pipeline that treats this as a synchronising feed: prior-year cases are revised, so a full re-ingest per release with versioning is the only approach that will not silently accumulate errors.

Licence

Freedom House publishes for public use and expects its work to be cited and circulated; attribution to the organisation and the specific report edition is the baseline obligation and is also self-protective, because these findings are politically contested and provenance is what makes them defensible. Specific reuse terms are stated on the publications and have not been uniform across products and years, so read the notice on the item you are using rather than generalising from another Freedom House product. Commercial redistribution — incorporating the case data into a paid risk or screening product — is the use most likely to need explicit permission, and asking is straightforward. There is a separate, non-legal obligation that matters more here than licence text: these records concern people who are being hunted, and republishing case detail in a form that increases their exposure is a harm regardless of what the licence permits.

Rate limits and fair use

No API and no reason for automated access beyond an annual retrieval. If you monitor the site for new publications, do so at a human cadence with an identifying user agent. Repeated scraping of a human rights organisation's publications infrastructure achieves nothing that one download a year would not.

Licensing changes, and it changes without warning. A dataset that was free for research this year may not be free for commercial or evidential use next year. Confirm the current terms before you build a dependency on it, and record the terms you relied on alongside the data — the licence in force at the time of collection is part of the provenance.

Collecting it

How Freedom House Transnational Repression is actually pulled, in the order you would set it up. Prefer the bulk or export interface over per-item lookups wherever one exists: it is kinder to the publisher, faster for you, and gives a reproducible snapshot rather than a series of point-in-time answers you cannot reconstruct later.

Method	Format	Cadence	Notes
Annual report and case list retrieval	CSV	annually, on publication	The authoritative artefact. Take the narrative report and the case data together and archive both; the methodology in the report is not recoverable from the data file.
Methodology capture	HTML	annually	Archive the methodology section for each release. The definitional boundary between counted and uncounted tactics has been elaborated over time, and comparisons across releases depend on knowing what changed.
Companion product retrieval	HTML	annually	Pull Freedom in the World and Freedom on the Net country assessments for the origin and host states in your scope. They supply the domestic governance context that explains the cross-border behaviour.
Primary source follow-through	HTML	per case	For any case that carries weight in your work, retrieve the underlying reporting, court records and government statements cited. The case entry is an index to evidence, not the evidence.
Cross-organisation reconciliation	HTML	periodic	Compare against specialist trackers covering particular regions, communities or tactics. Each organisation's case list is a partial view, and the union is materially larger than any one of them.

Ingesting it into the platform

Every step below is idempotent and cursor-based: interrupt one and it resumes from where it stopped rather than duplicating rows or losing progress. Collection is recorded per source, so a feed that quietly stops publishing shows up as a stale timestamp instead of silently thinning your coverage.

Register as an annual research artefact — Record the source in sources.php as an annual bulk publication with revision-in-place semantics, so that analysts understand both the lag and the fact that historical counts move between releases.
Import with release versioning — Load each release through import.php as a distinct version rather than merging into a running table, so that a case added or reclassified in a later edition is visible as a change rather than appearing to have always been there.
Model the origin-host pair as the primary relationship — During ingest.php, create the state-pair relationship as a first-class object rather than treating origin and host as attributes of an incident. The pair is the analytical unit that reveals permissive host environments.
Handle personal data restrictively — Where cases name individuals, apply the strictest handling class available in the platform, restrict access to the case, and exclude those records from any bulk export or sharing configuration by default rather than by exception.
Attach governance context to both states — Link origin and host records to country.php and country-risk.php with the relevant governance assessments, so that a case sits inside the legal and political environment that produced it rather than as a bare incident.
Correlate with policing and legal channels — Use correlate.php to align cases with extradition activity, international policing notices and asylum decisions where those are recorded, since the mechanism of a detention is frequently the actionable part.
Build the campaign view — Aggregate incidents by origin state and targeted community in campaign.php to distinguish sustained campaigns from isolated acts, which is the distinction that supports a pattern-of-conduct argument.
Route to human-rights and case workflow — Surface the results through human-rights.php and attach relevant cases in cases.php with the referral and protection pathway recorded alongside the analysis, so the record cannot be read without the duty-of-care context.

Registered sources and their last-collected state are listed in sources.php, and the scheduled chain that keeps them current is in automation.php.

How it is wrong, and how to tell

Every dataset is wrong in characteristic ways. Knowing which ways is the difference between using a source and being used by one, and it is the part of source evaluation most often skipped because it is the part that takes work.

The evidentiary standard is high and deliberately conservative. Cases are documented from public reporting, court records and NGO documentation, attribution to an origin state is required rather than assumed, and the methodology is stated plainly enough that you can see where the boundaries are drawn. The organisation's decision to count only direct physical acts is the strongest signal of its seriousness: it produces a smaller number that is defensible rather than a larger number that is not, and it accepts the resulting under-statement openly. Judge the source accordingly — as a rigorous floor. The weaknesses are documentation-driven rather than analytical: recent years are always incomplete, deniable operations are under-attributed, and coverage tracks the presence of people able to report. The political contestation of the publisher is a separate matter from the accuracy of the cases, and the appropriate response is corroboration rather than discount: for any case that carries weight, go to the cited primary sources and to a second organisation's documentation. Where those agree, the finding is solid. Where only Freedom House has it, say so in your product.

Characteristic false positives

A recent-year decline that is a documentation artefact. Cases surface months or years after the fact, so the most recent year in any release is under-counted, and reporting that as a decrease in transnational repression is the most common error made with this data.
Attribution to a state where the act was carried out by proxies with unclear direction. Criminal intermediaries, diaspora informants and unaffiliated sympathisers blur the line between state-directed and state-encouraged, and the resulting attributions are the most contestable entries in the database.
Host-country counts read as a ranking of danger. A country with many recorded incidents may simply have functioning police, media and courts that document them; the safest-looking host countries are sometimes the least transparent.
Incident counts read as victim counts. One operation against a group counts once, so origin-state comparisons are sensitive to how operations are structured rather than to the number of people harmed.
Lawful deportations coded as repression, and unlawful ones missed. The boundary depends on whether return exposes the person to persecution, which is a judgement made on incomplete information and made differently by different documenters.
The absence of a state from the database read as an absence of behaviour. States with small exile populations, effective deniability, or host environments where nobody reports will appear clean, and the dataset cannot distinguish those from restraint.
Duplicate accounting across releases. Cases are sometimes merged, split or reclassified between editions, so a naive union of two years' files produces both duplicates and phantom trends.
Narrative details treated as established fact. The case account reflects what the cited reporting said, which in contested incidents is frequently disputed, and the database does not adjudicate between competing accounts.

None of these make the source unusable. They make it a source that requires corroboration before an assertion built on it goes into a product, which is true of every source and admitted by few.

Ageing

Cases are historical records and remain true, but almost everything an analyst wants to do with them depends on context that moves. The origin state's capability and intent change with its leadership and with its security service structure; a pattern established under one administration is a weak guide under another, in either direction. Host-country behaviour changes faster still, and usually in response to exactly the exposure this dataset creates: several states have introduced foreign interference legislation, tightened extradition scrutiny or created reporting mechanisms since the series began, which changes both the risk and the likelihood of documentation. The counts themselves age backwards as well as forwards — a year's figure in a three-year-old file is lower than the same year's figure in the current release. A stale usage looks like a diaspora risk assessment resting on a host country's permissiveness that has since been legislated against, or a headline count quoted from an archived file that the publisher has since revised upward. Re-retrieve the current release whenever a number matters, and re-derive the host-country legal position from primary law rather than from the case record.

What this source feeds

A source is only worth what it lets you conclude. These are the disciplines that collect through it, the mission domains it serves and the data points it yields — every one is a tag, so you can follow any thread from here into the rest of the library.

Collected by these intelligence disciplines

Serves these mission domains

Transnational Repression

Yields these data points

How each sector uses Freedom House Transnational Repression

The same dataset is worked very differently depending on who you are, what authority you hold, and what you are ultimately producing. A military analyst is supporting a commander’s decision; a journalist is meeting a publication standard; an NGO caseworker is protecting a person. The records are shared — the constraints, thresholds and outputs are not.

🎖 Military and defence

Relevance is narrower here than for other sources in this category, and it is mostly about the security environment around people your organisation is responsible for. Where forces host, employ or protect nationals of states that practise transnational repression — interpreters, local staff resettled after operations, defectors, liaison contacts — this dataset defines a threat that persists after relocation and that host-country protection frameworks frequently do not cover. It is also relevant to understanding an adversary state's willingness to conduct operations on third-country territory, which is a capability and intent indicator in its own right. Handle any list of individuals in this category as protected information: a compilation of relocated persons is exactly the collection an origin state would most like to obtain, and the aggregation itself creates a risk that the individual records did not.

🕵 National intelligence

Transnational repression sits at the intersection of human rights reporting and foreign interference, and this is the best open baseline for the latter. Use it to characterise which states conduct operations on foreign territory against their own nationals, what methods they favour, how those methods vary by host country, and how the pattern has changed as host states have legislated. The origin-host pair is the analytically productive unit: it exposes which bilateral relationships enable rendition and which do not, which is a question about security cooperation agreements and intelligence liaison rather than about human rights. The dataset's floor-not-estimate character is the key methodological point for assessment writing — it supports statements about what is documented and cannot support prevalence claims. Expect the source to be contested on provenance by the states it names, and corroborate accordingly.

👮 Law enforcement

In most jurisdictions, several of the acts documented here are serious crimes committed on your territory by or on behalf of a foreign state: assault, abduction, unlawful detention, conspiracy, and increasingly specific foreign interference offences. This dataset is the open baseline for the threat, and for many forces it is the first indication that a community in their area is being targeted at all. Two operational implications follow. First, victims of transnational repression are structurally reluctant to report — immigration precarity, fear for relatives at home, and prior experience of police in the origin state all cut against it — so a low complaint volume is not evidence of a low threat, and community engagement has to precede case generation. Second, requests from the origin state for assistance, extradition or information about the same individuals are part of the same phenomenon, and cases documented here have repeatedly involved international policing channels being used against the target. Cross-check inbound requests against this record before acting on them.

🔍 Private investigation and corporate security

Private investigators and corporate security encounter this from two directions and one of them is a professional integrity problem. The legitimate direction is protective: advising clients who are exiles, dissidents or their employers on a threat model that includes surveillance, harassment, legal harassment and physical risk directed by a state. The illegitimate direction is being retained, directly or through intermediaries, to locate or profile someone who turns out to be a target of an origin state. This has happened repeatedly and the intermediary structures are designed to obscure it. Treat any instruction to locate a foreign national, particularly a journalist, activist or former official, as requiring know-your-client diligence beyond the ordinary standard, and use this dataset to recognise the pattern. A locate that ends in a rendition is a career-ending event at best.

📰 Journalism and OSINT media

For reporting, this is the dataset that makes a pattern visible behind incidents that were previously covered as isolated stories, and it gives editors a comparative frame: this origin state, this host country, this many documented cases, this method. The strongest reporting from it tends to be host-country accountability journalism — what the police did when a case was reported, whether prosecutions followed, whether deportation decisions complied with non-refoulement obligations. Handle the recent-year figures carefully; the current year is always under-counted and describing it as a decline is a predictable error. Source protection here is not routine care but a life-safety matter: exiles who speak to you are frequently the same people the origin state is looking for, and their relatives at home are the pressure point. Discuss publication risk with the person before you write, not after.

🌍 NGO, humanitarian and human rights

For human rights and refugee organisations this is core documentation and it is used exactly as intended: in submissions to UN special procedures, in asylum and protection cases, in litigation, and in advocacy for host-country legal reform. The case narratives support pattern-of-conduct arguments that individual testimony alone cannot, and the origin-host structure is directly usable in advocacy aimed at the host government rather than the perpetrator. For operational work with diaspora communities, the analysis of uncounted tactics matters more than the case count: coercion by proxy and digital harassment affect far more people than physical incidents, and a protection programme designed around the counted cases will address the smaller problem. Contribute back where you can document a case, and handle individual identities with the same conservatism the publisher does, which is more conservative than most casework norms.

🎓 University and research

The database has become a standard reference in a research area that barely existed a decade ago, and it supports work in comparative authoritarianism, diaspora studies, migration and refugee law, and international relations. The methodological facts to internalise are that the unit is a documented incident rather than a person, that the count is an explicit floor restricted to direct physical acts, that documentation lags events so recent years are censored, and that detection depends on host-country press freedom and diaspora organisation. Any quantitative design has to model that selection rather than assume it away, and the honest studies do so explicitly. Ethically, this is research on identifiable people at ongoing risk: institutional review is warranted, and the standard of care around re-identification and publication should exceed what a public dataset would normally require, because the adversary is a state with an active interest in the same records.

Playbook: working Freedom House Transnational Repression end to end

A repeatable sequence from first pull to finished product. Each phase states what you are trying to establish, not merely what to click — the objective is a defensible chain of reasoning, not a completed checklist.

Phase 1 — Read the methodology before the cases

Establish exactly what the database counts and what it deliberately excludes. Everything that goes wrong in the use of this source begins with treating a count of direct physical acts as a measure of transnational repression. Note the definitional boundary and quote it in whatever you produce.

Phase 2 — Retrieve the current release, not your archived copy

Cases are added and revised retrospectively, so a year's figure in an older file is lower than in the current one. If a number will appear in a product, take it from the release you can cite today and record the release identifier alongside it.

Phase 3 — Build the origin-host matrix first

Before looking at any individual case, tabulate origin states against host countries. This is where the analytically useful structure lives: which regimes act abroad, where they succeed, and which host environments are permissive. It also immediately exposes where the documentation is, which is the honesty check on everything that follows.

Phase 4 — Separate documentation capacity from behaviour

For each host country in your scope, characterise press freedom, diaspora organisation and the existence of a reporting mechanism. A country with few cases and no independent documentation is an unknown, not a safe one, and stating that explicitly is what separates an assessment from a chart.

Phase 5 — Follow the mechanism, not just the act

For each case, establish how it was accomplished: international policing channels, an extradition request, a deportation decision, a covert operation, a proxy. The mechanism determines who else was involved, which institutions failed, and where an intervention is possible — and it is usually in the narrative rather than the tabulated fields.

Phase 6 — Corroborate from the cited primary sources

For any case that will carry weight, retrieve the underlying court records, reporting and government statements. The database is an index to evidence and is politically contested; a product built on the index alone will be attacked on provenance and will have no second line of defence.

Phase 7 — Widen to the uncounted tactics

Deliberately extend the picture with coercion by proxy, mobility controls, abusive legal process and digital harassment using specialist sources. For most targeted communities these are the dominant experience, and a protection or policy product scoped to physical incidents will miss what most people are actually facing.

Phase 8 — Assess the host state's response

Establish what happened after each incident: investigation, prosecution, diplomatic action, or nothing. The database is thin here and the answer is usually available from court and press records. This is the variable policy turns on, and it is where original analytical value is easiest to add.

Phase 9 — Turn analysis into protection, not into a list

Where the work touches real people, convert findings into protective advice, referral pathways and host-institution engagement. Do not compile or retain a roster of at-risk individuals unless there is a specific protective purpose, a lawful basis and access controls that would survive a breach. The compilation is itself a risk to the people in it.

Phase 10 — Track the legal and policy response over time

Monitor foreign interference legislation, extradition reform and international policing accountability mechanisms in the host countries you cover. The environment is changing quickly, and a permissiveness judgement from three years ago is frequently obsolete in one direction or the other.

Phase 11 — State the floor explicitly in the product

Whenever a count appears in your output, attach the sentence that says what it excludes and that recent years are under-counted. Reviewers, journalists and opposing counsel will find that caveat if you do not state it, and stating it costs nothing while omitting it costs the whole argument.

Phase 12 — Record with restricted handling

File the analysis in cases.php and human-rights.php with individual identities under restricted access, excluded from default exports and sharing. The origin states named in this dataset conduct intelligence operations to obtain exactly this material, and your handling posture should assume that.

The platform ships this as a step-checked workflow in playbooks.php, so progress is recorded against a case rather than held in someone’s head.

What to pair it with

No single source carries a finding. These are the datasets that corroborate, extend or contradict this one — and a source that contradicts is worth more than one that agrees, because it is the only thing that will tell you when you are wrong.

Source	Relationship	What it adds
Safeguard Defenders	extends	Specialist documentation of overseas policing operations, involuntary returns and coercion by proxy, with deep coverage of tactics the case database excludes.
Fair Trials	extends	Casework and analysis on the abuse of international policing channels and extradition processes, which is the mechanism behind a substantial share of documented detentions.
INTERPOL	prerequisite	The organisation whose notices and diffusions are misused in these cases, and whose file review commission is the formal challenge route. Understanding the process is necessary to assess any case involving a notice.
Committee to Protect Journalists	corroborates	Case-level documentation of journalists targeted in exile, with independent sourcing that frequently corroborates or extends individual cases.
Reporters Without Borders	corroborates	Press freedom casework covering exiled journalists and the pressure applied to them and their families across borders.
Human Rights Watch	corroborates	Detailed country and incident investigations that frequently constitute the underlying documentation for cases in this database.
Amnesty International	corroborates	Independent documentation, urgent action casework and the security research that covers the digital dimension of the same campaigns.
The Citizen Lab	extends	Forensic evidence of the digital component — spyware deployed against exiles and their contacts — which is documented in the analysis here but not counted in the case database.
Freedom in the World	prerequisite	The same publisher's country assessments, which supply the domestic governance context that makes an origin state's external behaviour interpretable.

Legal, ethical and operational constraints

This dataset concerns identifiable individuals who are subject to an ongoing threat from a state, which places it at the most sensitive end of the personal data spectrum. In most data protection regimes the records constitute special-category data because they reveal political opinion, ethnicity or religious affiliation, and processing requires both a lawful basis and a demonstrable necessity — a general research interest is unlikely to be sufficient for retaining a named-individual dataset. Storage, access control, retention limits and export restrictions should be set at the level you would apply to victim data, because that is what it is. Onward disclosure is the acute risk: sharing a case list with a partner organisation, a client or a foreign government could place people in danger, and the fact that the underlying material is public does not neutralise the harm of aggregation. There are also positive legal duties in play: in a growing number of jurisdictions, acts documented here constitute foreign interference or specific criminal offences, and organisations that become aware of a credible threat to a person may have reporting or duty-of-care obligations. Take a written position on both the protective and the processing questions before you build anything.

Operational security

Two exposures matter, and the second is the serious one. Reading and downloading public human rights reports is low risk in most environments, although in states that treat contact with foreign human rights organisations as a security matter, patterned access from inside the country is not. The serious exposure is what your work product becomes. An analysis of transnational repression is, viewed from the origin state's side, a compilation of who has left, who is speaking, where they are, and who is protecting them — precisely the intelligence requirement that drives the phenomenon in the first place. Origin states have conducted operations against the organisations and researchers who document this, including recruitment of insiders and technical intrusion. Handle your own working material as a target: restrict access, avoid consolidating identities into a single searchable store, keep diaspora contacts out of shared systems, and think carefully before naming host-country institutions and individuals who are providing protection. If you engage directly with targeted people, follow their operational preferences rather than your organisation's default communications, and never introduce them to each other without consent.

Two rules that hold regardless of jurisdiction. Collection that is lawful is not automatically proportionate, and a dataset assembled for one purpose does not carry consent for another. Where the records concern identifiable people, the question is not only whether you may hold the data but whether holding it serves the purpose you are accountable for.

Is it earning its place?

Sources accumulate. Feeds get added during an incident and are never reviewed again, and a decade later the pipeline is carrying dead weight that nobody dares remove. These are the measures that show whether Freedom House Transnational Repression is contributing anything, and they are worth baselining now so the answer is available later.

Share of cases in your working set for which you retrieved and archived at least one cited primary source, as the measure of whether your analysis can survive a provenance attack.
Coverage of the uncounted tactics in your assessments, tracked as the proportion of country products that address coercion by proxy and mobility controls alongside physical incidents.
Number of host countries in your scope for which you have characterised documentation capacity, so that low case counts can be interpreted rather than reported.
Currency of the host-country legal position in your assessments, measured as time since the extradition, deportation and foreign interference framework was last verified against primary law.
Count of products in which the floor caveat appears alongside the case count, treated as a defect rate when it is missing.
Referrals and protective actions generated per analytical product where real people are involved, because an assessment of transnational repression that produces no protective outcome has not been operationalised.
Access reviews completed on any store containing named individuals, at a frequency appropriate to data that a hostile state actively seeks.

Beware of volume. Indicator counts rise easily and say almost nothing. Unique contribution — findings this source produced that no other source in your stack would have — is the measure that matters, and it is usually far lower than anyone expects.

Tradecraft notes

The distinctions that separate a competent analyst from a fast one:

Treat the count as a floor and say so every time. The methodology deliberately excludes the most common tactics, and any argument that depends on the number being an estimate of scale is built on sand.
Never report the most recent year as a trend. Documentation lags events, so the newest figure is always the least complete, and the apparent decline is an artefact that recurs in every release.
The origin-host pair is the unit that matters. Origin states tell you about intent; host countries tell you about opportunity, and policy operates almost entirely on the second.
Read the mechanism out of the narrative. Whether a person was taken by covert operation, handed over through a policing channel or removed by an immigration decision changes who is accountable and what can be done, and none of that is in the categorical fields.
A quiet host country may be an opaque one. Before concluding that a jurisdiction is safe for exiles, check whether anyone there is in a position to document a case.
Corroborate before you name. These findings are contested by well-resourced states, and a product resting on a single source will be attacked on provenance rather than on substance.
Do not build a list of people. The aggregation is more dangerous than any individual record in it, and the intelligence requirement it satisfies belongs to the perpetrator.
Watch the inbound side. Requests for assistance, extradition and information from origin states are part of the same phenomenon, and this dataset is the reference for recognising when a lawful-looking request is an instrument of repression.
Follow the family. The pressure point in most of these campaigns is relatives who never left, and an analysis that stops at the border has missed the mechanism that actually controls the target.

Questions analysts actually ask

Why is the number of cases so low relative to what I hear about?

Because only direct physical incidents are counted. Threats against relatives at home, passport cancellations, abusive extradition requests and online harassment campaigns affect far more people and are analysed in the reports rather than tabulated. The count is an intentional floor, chosen for defensibility over completeness.

Does a fall in the most recent year mean the problem is easing?

Almost certainly not. Cases surface months or years after they occur, so the newest year in any release is systematically under-documented. Compare only years that have had several releases to mature, and say so when you publish.

Can I use this as a screening list for people at risk?

No. It is an incident record, not a register of targeted individuals, and building a roster of at-risk people creates exactly the compilation that origin states are trying to obtain. If you need to support individuals, work through protection organisations and referral pathways rather than through a database.

How is 'transnational repression' distinguished from ordinary law enforcement cooperation?

By whether the act is unlawful in the host country or violates protection obligations — an abduction, an assault, a detention without proper process, a removal to a country where the person faces persecution. Legitimate extradition under judicial supervision is not in scope; the abuse of extradition channels is.

Is Freedom House a credible source given its funding?

The methodology is conservative, the cases are sourced, and the findings have held up under contestation. The funding matters for two practical reasons: coverage of states aligned with its principal funder is thinner, and named states will attack the provenance. Corroborate load-bearing cases from primary sources and from a second organisation.

Which host countries are the most dangerous for exiles?

The database cannot answer that directly, because recorded incidents track documentation capacity as well as risk. A host country with a free press, organised diaspora and functioning police will show more cases than an opaque one where more may be happening. Interpret the distribution alongside a documentation-capacity assessment.

How should I handle the individual names in the cases?

As victim data. Restrict access, exclude from default exports, avoid consolidating identities into a searchable store, and do not republish detail that increases a person's exposure. Publication by someone else does not transfer the risk decision to you.

What is the role of international policing channels in these cases?

A substantial one. Notices and requests through those channels have repeatedly been the mechanism by which exiles are detained abroad, and the case narratives frequently record it. If your work involves inbound requests, cross-checking against this record before acting is a basic and increasingly expected control.

Does the dataset record what host governments did about it?

Largely not, and that is the most useful gap to fill yourself. Investigation, prosecution, diplomatic response or inaction is usually recoverable from court and press records, and it is the variable that accountability work and policy engagement actually operate on.

Standards, formats and interoperability

What this source speaks natively, and what it has to be translated into before a partner can consume it. Work that arrives in a recognised format is easier to defend, easier to hand over and easier to automate against:

The non-refoulement obligation under the Refugee Convention and the Convention against Torture is the legal test that distinguishes an unlawful removal from a lawful deportation, and it governs how the deportation category should be read.
The International Covenant on Civil and Political Rights, including the extraterritorial application of its obligations, is the framework under which an origin state's conduct abroad is assessed.
UN special procedures — special rapporteur communications and working group opinions — are the international mechanism these cases are most commonly escalated to, and their case numbering provides an independent reference for corroboration.
International policing notice and diffusion procedures, and the file review mechanism that exists to challenge them, are the process layer behind a significant share of documented detentions.
Foreign interference and agent registration legislation in a growing number of host jurisdictions is the domestic legal frame under which these acts are now investigated, and it varies enough that host-by-host assessment is required.
ISO 3166 country coding is required to join origin and host states to the rest of the platform's geography; the published data uses names and normalisation is an ingest responsibility.
The platform treats named individuals in these records under restricted handling, excluded from default STIX, MISP and CSV exports, because sharing configurations designed for indicators are inappropriate for victim data.

References

Primary documentation and authoritative references for this source. Publishers revise and retire material, so treat the retrieval date as part of the citation and re-check before relying on any of it in a formal product.

Transnational Repression — Freedom House. The report series and case database home, including methodology and the analysis of tactics excluded from the count.
Freedom House — Freedom House. The organisation and its wider country assessment products, which supply the governance context for both origin and host states.
Freedom in the World — Freedom House. Annual country assessments of political rights and civil liberties, the domestic baseline against which cross-border behaviour is interpreted.
Safeguard Defenders — Safeguard Defenders. Specialist documentation of overseas policing, involuntary returns and coercion by proxy — the uncounted tactics, covered in depth.
Fair Trials — Fair Trials. Analysis and casework on the abuse of international policing and extradition mechanisms, the process route behind many detentions.
INTERPOL — INTERPOL. The notice and diffusion system, its stated safeguards and the file review commission that provides the formal challenge route.
Committee to Protect Journalists — CPJ. Independent case documentation for exiled journalists, frequently corroborating or extending individual entries.
Human Rights Watch — Human Rights Watch. In-depth incident and country investigations that often constitute the underlying documentation for a case.
Amnesty International — Amnesty International. Independent documentation and urgent action casework, plus the security research covering the digital dimension of the same campaigns.
The Citizen Lab — University of Toronto. Forensic evidence of spyware deployed against exiles, the technical counterpart to the physical incidents counted here.
Office of the United Nations High Commissioner for Human Rights — OHCHR. Special procedures communications and the international legal framework, including the extraterritorial reach of state obligations.

Link integrity: every reference above was verified with a live request when this page was generated. Where a publisher had moved or withdrawn a document, the link was repointed at a preserved copy in the Internet Archive and marked as archived. Anything with no reachable copy anywhere had its link removed rather than left to rot — the source is still credited, it simply cannot be linked.

Put it into practice

The Quantus Intel threat intelligence platform operationalises this source: it models each incident as an origin-state to host-country relationship rather than a bare event, carries the documented-floor caveat through to every derived count, links origin and host states to their governance assessments, and holds named individuals under restricted handling that is excluded from sharing and export by default.. Browse the full source catalogue, or follow any tag above into the rest of the library.