InSight Crime: Intelligence Source Guide

InSight Crime is a non-profit newsroom and think tank covering organised crime in Latin America and the Caribbean, built on field reporting, court records and interviews. Its group profiles and country dossiers are the most widely reused open descriptions of criminal structures in the hemisphere – which is exactly why you should know how they were made.

At a glance

Source	InSight Crime
Category	Conflict, Crime & Human Security › Organised Crime, Gangs & Piracy
Homepage	https://insightcrime.org/
Format	`HTML`
Access	Open — no account required
Disciplines	Criminal Intelligence, Open Source Intelligence
Mission domains	Gangs & Street Crime, Organized Crime

Organized-crime & gang investigative reporting (Americas). — as catalogued in the platform’s own source registry.

InSight Crime publishes investigative journalism and analysis about organised crime in the Americas. The output falls into four recognisable types. There are investigations: multi-part pieces built from field reporting, interviews with police, prosecutors, victims and members of criminal groups, and documentary material from court files and government records. There are criminal group profiles: standing reference pages on named organisations describing origin, leadership, structure, territory, criminal economies and allies or rivals. There are country profiles doing the same job for national criminal landscapes, including institutional and political dimensions. And there are recurring data-flavoured products, most prominently an annual compilation of homicide figures across the region drawn from official national sources with methodological notes about each. Everything is published as web articles in English and Spanish, tagged by country and by criminal actor, with no structured API and no dataset download. What you extract from it are named people, named organisations, named places, dated events and asserted relationships – all of them in prose.

Its analytical job is structural description, which is the thing threat data does worst. Indicator feeds, seizure statistics and homicide counts tell you that something is happening; they do not tell you which organisation controls a corridor, who split from whom in which year, which faction now runs extortion in a given neighbourhood, or how a group's revenue mix shifted from drug transport to migrant smuggling to illegal mining. InSight Crime does that work by sending reporters to places and reading the case files, and it does it consistently enough across two decades that you can trace an organisation's evolution through their archive. For CRIMINT work in the Americas it is often the only open description of a group that goes beyond a law enforcement press release. The corollary is that it is also, for many groups, the origin of the description everyone else uses – which means corroborating a claim by finding it repeated in three other outlets frequently means finding the same InSight Crime article three times.

Who publishes it, and why that matters

InSight Crime is a non-profit organisation. It was founded around 2010 by journalists working on organised crime in Colombia and the wider region, it has had institutional homes in the US academic sector and operational bases in Latin America, and it is funded by philanthropic foundations and by grants including from government donors that fund transparency, rule-of-law and security programming. That funding model has three implications you should hold in mind. Coverage follows funding: regions and themes with active grant programmes get sustained investigation, and countries without a donor interest get thinner treatment regardless of how much crime is there. Government funding of any kind attracts the accusation of alignment, and in polarised environments the organisation's work is attacked on that basis by governments it criticises; the accusation is not evidence of bias, but the funding relationships are public and you should be able to state them if a client challenges a claim. And non-profit newsrooms are structurally fragile – staffing, language coverage and regional depth change with grant cycles, so an area that was covered exhaustively three years ago may be covered thinly now, and that change is invisible in the archive.

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
`article title and dek`	string	The headline and standfirst, which frequently carry the substantive claim in compressed form. Useful for triage, dangerous as a citation, because a headline claim is often hedged in the body.	Keyword and entity extraction for monitoring; the claim to verify against the article body.
`publication date`	timestamp	When the piece was published. For standing group and country profiles this is the last significant update rather than the date of the underlying reporting, and the two can be years apart.	Timeline placement; always check whether a profile's date reflects a refresh or the original writing.
`country tags`	array	The countries a piece concerns, applied editorially. They reflect where the story is set rather than every jurisdiction implicated in the network described.	Country dashboards and regional filtering; supplement with entity extraction because transit and destination countries are often untagged.
`criminal actor tags`	array	Named groups the piece concerns, drawn from the site's own taxonomy of organisations. This taxonomy is an editorial construct, not a legal or official list, and names in it may not correspond to any indicted entity.	Actor profiles; the join between narrative reporting and any structured actor model you maintain.
`person names in body`	string	Individuals named in the reporting: alleged leaders, indicted figures, officials, victims and sources. Legal status varies enormously and the text usually states it, though downstream reuse strips that qualification.	Sanctions and designation lists, court dockets, corporate registries and asset records – always after confirming the exact legal status asserted.
`alias and nickname`	string	Criminal aliases, which in this region are frequently the primary identity in both street usage and prosecution documents. Aliases are reused across individuals and regions and are a notorious source of conflation.	Court records and wanted notices, but only when combined with at least one other identifier such as a date of birth, a charge or a place.
`location references`	string	Municipalities, neighbourhoods, corridors, border crossings and ports named in the text. Often precise to the neighbourhood, which is finer than most structured sources on the region achieve.	Geographic entities; mapping of territorial control claims, which should always be plotted as claims rather than as facts.
`criminal economy references`	enum	The revenue lines described: cocaine transport, extortion, illegal mining, migrant smuggling, human trafficking, fuel theft, timber, wildlife, arms. This is the most consistently useful structured attribute you can extract.	Financial crime and commodity-flow analysis; the bridge from a group profile to trade, seizure and sanctions data.
`relationship assertions`	string	Statements that one group split from, allied with, absorbed, fought or paid another, or that an official protected one. These carry the analytical weight of the source and vary widely in evidentiary support.	Link analysis, but only with the sourcing basis attached to each edge – an alliance asserted by a prosecutor and one observed by a reporter are not the same edge.
`sourcing basis`	string	How the piece knows what it claims: court documents, official statements, interviews with named or anonymous figures, or the outlet's own field observation. Usually stated in the text and almost always discarded when the claim is reused.	Confidence weighting; this is the field that should determine how much of your assessment a claim can carry.
`homicide and statistical figures`	int	Numbers drawn from official national and municipal sources in the annual tallies and in individual pieces, with the originating institution normally named. The figures are as good as those institutions and the notes usually say so.	National crime statistics, and comparison against independent observatories; never treat the compilation as a primary count.
`language edition`	enum	English or Spanish. The two editions are not always identical, publication timing differs, and some material appears in only one – which matters if you are monitoring for a specific development.	Coverage completeness checks; monitor both editions if the subject is regionally significant.

Coverage — and what is not in it

Latin America and the Caribbean, with the deepest and longest-running coverage of Colombia, Mexico, Venezuela, the Northern Triangle of Central America, and Brazil, and periodic coverage extending across the Andean and Southern Cone countries, the Caribbean basin, and the criminal economies that link the region to the United States, Europe and West Africa. The thematic range covers the drug trade end to end, extortion and local racketeering, human smuggling and trafficking, environmental crime including illegal mining, logging and fishing, arms trafficking, prison-based organisation, and the corruption and state-complicity dimension that ties them together. Cadence is that of a newsroom: several pieces a week, with major investigations appearing in bursts and standing profiles updated irregularly. The critical coverage fact is that the archive is unevenly deep by both geography and time. A group that was the subject of a funded investigation has a rich, well-sourced dossier; a group of similar significance in a country without that attention may have one article from six years ago. Do not read absence of coverage as absence of activity, and do not read depth of coverage as proportionate to importance – it is proportionate to attention.

Known blind spots

Absence of evidence here is not evidence of absence. These are the conditions under which InSight Crime will not show you something that is nevertheless real:

Regional scope is a hard boundary. Networks that originate in the Americas but operate in Europe, West Africa or Asia are covered only where they touch the region, so the far end of a trafficking chain is usually missing.
Coverage follows funding and staffing rather than criminal significance, so a lightly covered country tells you about the newsroom's grant portfolio and not about the state of organised crime there.
Prison, rural and closed environments are described largely through the accounts of officials and of people who got out, which systematically over-represents the state's framing of internal structure.
Groups too small, too local or too new to have attracted prosecution or press attention are invisible, even where they dominate violence in a specific municipality.
Anything a source could not safely say is absent. Reporting on organised crime in the region carries lethal risk to journalists and interviewees, and the practical result is that some networks and some official complicity are under-described precisely where the danger is greatest.
State-linked criminality is harder to report than non-state criminality, so the balance of the archive tilts towards named non-state groups and away from the officials, businesses and security forces that enable them.
There is no structured dataset behind the prose. Territorial control maps, leadership charts and network diagrams are editorial renderings of reporting, not the output of a coded dataset you can audit or reproduce.
The annual homicide compilation inherits every weakness of the official statistics it aggregates, including definitional differences between countries, changes in counting rules, disappearances excluded from homicide counts, and political pressure on the figures.
Currency decays silently on standing profiles. A group profile carries an authoritative tone whether it was refreshed last month or last decade, and nothing in the layout tells the reader which.

Write the blind spot into the product. A statement that something “was not observed in InSight Crime” 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 site is free to read, in English and Spanish, with no paywall and no registration for the main article archive. There is no public API and no dataset export; collection means either reading it or parsing HTML and any syndication feed the site publishes. For an intelligence operation the realistic pattern is monitoring rather than bulk ingestion: subscribe to the newsletter or feed, filter on your countries and actors of interest, and pull individual pieces into the case record when they bear on live work. Bulk-harvesting the archive is technically straightforward and analytically less useful than it appears, because the value is in the sourcing and nuance of specific pieces rather than in aggregate text. If you are considering republication or a formal partnership, contact the organisation – non-profit newsrooms generally welcome the reuse that credits them and are firm about the reuse that does not, and the terms are worth establishing in writing rather than inferring from a footer.

Licence

Treat the content as copyrighted journalism unless a specific piece says otherwise. Some non-profit newsrooms permit republication under stated conditions and some do not, and the arrangement can vary by piece and by partnership, so check the terms on the site and on the individual article rather than assuming a blanket permission. What is safe in any case is normal journalistic and analytical practice: read it, quote briefly with attribution, cite the specific article and date, link to it, and build your own analysis on top. What is not safe is lifting profiles wholesale into a commercial product, mirroring the archive, or presenting their reporting as your own collection. If your product will assert something significant that rests entirely on their work, say so explicitly – both because it is honest and because it tells your reader exactly how much independent corroboration the claim has, which is the thing they most need to know.

Rate limits and fair use

There are no published limits, which means the constraint is courtesy toward a non-profit with a modest hosting budget. Poll any feed at a normal reader's frequency – hourly at most – and never crawl the archive aggressively. Identify your collector honestly. If you need the archive at scale for research, ask them; a request is more likely to succeed and far less likely to get you blocked than a crawler that takes their site down. Cache what you retrieve, do not re-fetch unchanged pages, and treat any error response as a signal to back off rather than to retry harder.

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 InSight Crime 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
Article monitoring by tag	HTML	daily	Filter the stream on your countries and criminal actors and route matches to an analyst. This is the highest-value collection mode and it costs almost nothing.
Standing group and country profiles	HTML	check quarterly	Pull the profiles for actors in your portfolio and diff them against your stored copy. A quiet edit to a profile is often the first open signal that a leadership or territorial claim has changed.
Annual homicide compilation	HTML	annual	The regional tally with per-country methodological notes. Ingest the notes, not just the numbers, because the notes are what make the numbers comparable or not.
Major investigations	HTML	as published	Multi-part series carry the densest entity and relationship content in the archive. These warrant full manual extraction into the case rather than automated processing.
Manual analyst extraction	CSV	per case	For any claim entering an assessment, record the article, date, the exact claim and its stated sourcing basis. This is slower than scraping and it is the only method that preserves what makes the source worth using.

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 analytic reporting source — Record it in sources.php as narrative reporting rather than as an indicator feed, so nothing derived from it inherits feed-level confidence anywhere in the platform.
Ingest documents, not facts — Load articles through import.php as documents with title, date, URL, tags and full text. Resist the temptation to auto-derive entities into the graph at load time; the extraction step needs judgement that the loader does not have.
Extract entities with the sourcing basis attached — Person, organisation, location and criminal-economy mentions become candidate entities, each carrying the article reference and how the article knew it. An entity created here is a candidate for analyst promotion, not a confirmed node.
Model relationships as claims — Every asserted alliance, split, control or protection relationship enters link-analysis.php as a claim with an author, a date and a basis. If two articles assert contradictory relationships, both edges exist and the contradiction is visible rather than resolved silently.
Reconcile aliases carefully — Route alias-based identities through an analyst before merging them into a person entity. Alias collision is the dominant identity error in this region and an automated merge will confidently fuse two different people.
Cross-reference to official records — Check named individuals and organisations against sanctions designations, wanted notices and court dockets, and record which claims have official corroboration and which rest on the reporting alone. This produces the corroboration state that should drive confidence in organized-crime.php.
Summarise, never generate — Copilot summarisation over ingested articles is appropriate and useful. No entity, relationship or attribution in the platform should originate from a model reading this source – every node traces to a sentence in a published article with a citation.
Diff standing profiles on a schedule — Re-collect group and country profiles periodically and store the diff. Changes to a profile are a low-noise signal and are far more informative than another article in the stream.

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.

As journalism, it is strong: reported from the ground, generally careful about attributing claims, explicit about legal status, and willing to publish uncertainty. As an intelligence input, its quality is entirely a function of how you read it. The right test is to take a claim you can check independently – an indictment, a sanctions designation, a court judgment – and trace how the reporting characterised it before and after the official action. You will usually find that the substantive picture was right, the timing was early, the leadership names were the ones prosecutors later used, and the organisational chart was more definite in the prose than the evidence strictly supported. That last point is the one that matters. Criminal organisations in the region are more fluid, more federated and less hierarchical than any narrative description can convey, and the pressure of writing readable prose pushes every outlet, including this one, towards cleaner structures than reality contains. Use the reporting for what is happening and who is involved; be sceptical of any diagram that shows a clean chain of command; and always propagate the sourcing basis, because a claim from a court filing and a claim from an anonymous interview should not carry equal weight three documents downstream.

Characteristic false positives

Structure imposed on fluidity. Named groups with boxes and arrows read as stable organisations, while the underlying reality is frequently a shifting set of local cells using a common brand, and the diagram outlives the arrangement it described.
Prosecutorial framing adopted wholesale. Where reporting rests on indictments and official statements, it inherits the state's theory of the organisation, which is constructed to support charges rather than to describe a network accurately.
Brand names treated as entities. A label used by a prosecutor, a rival, the press and eventually by the group itself does not necessarily denote a single organisation, and treating it as one creates a phantom actor that then acquires attributes from every article that mentions it.
Alias conflation. Nicknames repeat across individuals and regions, and a person entity assembled from alias matches across several articles will often turn out to be two or three people.
Circular corroboration. A claim originating here is picked up by regional and international press and returns as apparent independent confirmation. Always trace a claim back to its first appearance before counting sources.
Standing profile staleness. A profile written at the height of a group's power reads as current years after the group fragmented, and readers take the confident present tense at face value.
Official statistics inherited without their caveats. The homicide compilation is only as good as each country's counting rules, and the per-country notes explaining why a figure is not comparable are usually stripped when the number is reused.
Absence read as significance. A country or group with little coverage looks unimportant, when it may simply have fallen outside the newsroom's current funding and staffing footprint.

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

News articles age as reporting always does: the facts as of the publication date remain the facts as of that date, and their relevance decays at the speed of the underlying conflict, which in this region can be weeks. Standing profiles age differently and more dangerously, because their form implies currency. A group profile has no visible decay indicator, and a reader arriving from a search engine has no way to tell whether the leadership section describes the current situation or the situation before a capture, an extradition or a split. Assume any profile is stale until you have checked the date and cross-checked the leadership against recent news and official actions. Relationship claims age fastest of all: alliances between criminal groups in the region routinely last months, and an alliance edge in your graph that came from a two-year-old article is worse than no edge, because it will be treated as background rather than as a dated claim. The mitigation is structural – date every claim at ingest, decay relationship confidence explicitly with time, and re-verify any edge before it appears in a product.

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

Yields these data points

How each sector uses InSight Crime

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

For forces engaged in support to civil authorities, border security or countering transnational criminal organisations in the hemisphere, this is the open baseline on who operates where and how they generate revenue. It is particularly useful for understanding the interaction between criminal economies and armed groups in the border and jungle areas where the distinction between insurgency and criminality has collapsed. Read it for the human and economic terrain rather than for order of battle: it will tell you which economies fund which actors and where the friction points are, and it will not give you strengths, dispositions or capabilities at any usable resolution. Treat every organisational chart as an editorial construct and validate territorial claims against your own collection before they inform any posture decision.

🕵 National intelligence

For strategic and warning work on the Americas, the value is in structural change over time. Reading a group's coverage chronologically shows fragmentation, generational leadership turnover, and shifts in revenue mix from drugs to mining to migrant smuggling, and those shifts precede changes in violence patterns and in the state's relationship with the group. The relationship claims support competing-hypothesis work about who benefits from a killing or a corridor closure. Because the outlet is often first to a development, it is also a useful tipper: an unexplained profile update or an investigation into an unfamiliar group is worth an analyst's attention before it becomes an official reporting item.

👮 Law enforcement

For investigators, this is lead material and context, never evidence. It can tell you that a network has been described operating in your jurisdiction, identify aliases and locations worth checking against your own holdings, and point you to the court records and official actions that are the real evidentiary sources. Use it to orient before liaison with a foreign counterpart, because it will tell you how the local landscape is understood there. Be disciplined about legal status: a person described as an alleged leader is an allegation, and importing that description into an intelligence record without the qualifier creates a downstream risk to both the case and the individual.

🔍 Private investigation and corporate security

For corporate security, due diligence and investigations work in the region, this is often the only open account of who actually controls a locality – which is the operative question for a mine, a plantation, a port terminal, a logistics route or a construction project. It supports extortion risk assessment, counterparty screening beyond the sanctions lists, and realistic security planning for personnel in specific municipalities. Use it to generate hypotheses and named leads, then verify through registries, court records and local counsel before anything enters a client deliverable, and be explicit about which findings rest on journalism rather than on record evidence.

📰 Journalism and OSINT media

For journalists, InSight Crime is a peer newsroom and should be treated as one: a source of leads, context and named contacts, and a body of work to build on and credit rather than to recycle. The specific discipline is provenance. Because so much regional and international coverage of Latin American organised crime traces back to this outlet, a claim that appears widely sourced frequently is not, and a story premised on consensus can collapse when the consensus turns out to be one article. Trace claims to their origin, contact the original reporters where a story warrants it, and be scrupulous about attributing analysis you did not do.

🌍 NGO, humanitarian and human rights

For humanitarian, human rights and development organisations, the reporting on extortion, forced displacement, forced recruitment, migrant smuggling and control of local populations is directly relevant to programme risk and to protection work. It tells you which actor governs a territory you intend to operate in, which is a different and more useful question than which state institution nominally administers it. Handle victim material with care: the reporting names people who have suffered serious harm, and reproducing those details in a programme document or a public report can put them at renewed risk. Aggregate for programming, and route protection concerns through established local mechanisms.

🎓 University and research

For researchers in criminology, security studies and Latin American politics, this is a substantial and unusually consistent body of qualitative material on criminal organisation, and it is used as such widely. Treat it as journalism in your methodology: cite specific articles rather than the organisation, acknowledge that coverage intensity is a function of the newsroom's resources rather than of the phenomenon, and be explicit when a claim about group structure rests on a single reported source. It is a poor substitute for a coded dataset and an excellent complement to one, particularly for generating hypotheses about mechanisms that quantitative sources can then test.

Playbook: working InSight Crime 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 — Define the entity before you search

Decide whether you are investigating a named organisation, a territory, a criminal economy or a person, because each requires a different route through the archive. Searching a group name first is the default and it is often the worst choice, since brand names attract material about several distinct structures. Starting from a place or an economy usually produces a cleaner picture.

Phase 2 — Read the archive chronologically

Pull everything on your subject and read it oldest first. This is the single highest-value habit with this source. Chronological reading exposes fragmentation, name changes, leadership turnover and the moments where the outlet's own understanding shifted – all of which are invisible if you read the most recent piece and stop.

Phase 3 — Separate claim from sourcing at the point of reading

For each substantive claim, write down what is claimed and how the article knows it: court document, official statement, named interview, anonymous source, or the reporter's observation. Do this as you read. Reconstructing it later is far more work and is usually skipped, which is how unsourced claims end up carrying assessments.

Phase 4 — Build the entity list before the network

Extract every named person, organisation, place and economy first, without drawing any connections. You are producing a vocabulary. Half the errors in this kind of work come from building a network diagram while still uncertain whether two names refer to the same thing.

Phase 5 — Resolve identity against official records

Take your named individuals to sanctions designations, wanted notices, court dockets and corporate registries. Each match converts a journalistic identity into a documented one and gives you a legal status you can state. Unmatched names stay flagged as reported-only, and any assessment resting on them says so.

Phase 6 — Plot control claims as claims

Where the reporting asserts territorial control, map it with the article, the date and the basis attached. Control in this region is contested, partial and reversible, and a map that renders claims as facts will be presented to a client as knowledge of who runs a place, which is the most consequential error you can make with this material.

Phase 7 — Characterise the revenue mix

For each actor, build the list of criminal economies attributed to it over time. Revenue mix is the most stable and most predictive attribute in the archive: it determines which routes and assets matter to a group, which other actors it must deal with, and where enforcement pressure will actually hurt. It also drives your pivot into financial-crime.php and commodity data.

Phase 8 — Test for circular sourcing

For each claim you intend to rely on, search for its earliest appearance and check whether the other sources citing it are independent or downstream. Regional organised-crime reporting is heavily recycled. A claim that survives this test can carry weight; one that does not should be attributed to a single outlet in your product.

Phase 9 — Corroborate with structured data

Take the claims that matter and test them against seizure statistics, homicide data, sanctions actions, prosecutions and trade flows. Where narrative and structured data agree, your confidence rises legitimately; where they disagree, the disagreement is usually the most interesting thing in the assessment and should be reported rather than smoothed.

Phase 10 — Date and decay every relationship

Assign every edge in your network a date and an explicit decay assumption based on how fast alliances shift in that theatre. Then look at the network with the stale edges removed. Analysts are routinely surprised by how much of their picture rests on relationships that were reported years ago and never re-verified.

Phase 11 — Localise before you conclude

Before finalising anything about a specific municipality, port or corridor, check local and national reporting in the language of the country. The regional outlet will have the structural picture and the local press will have the last six weeks, and the difference between them is often where your assessment actually lives.

Phase 12 — Write the provenance into the product

State plainly which findings rest on investigative journalism, which on official records, and which on your own collection. Clients and courts treat these differently and they are right to. A product that distinguishes them is more useful and far more defensible than one that presents a seamless narrative.

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
Global Initiative Against Transnational Organized Crime	corroborates	Comparable analytical depth with global rather than regional scope, plus structured resilience and criminality indexing that provides a comparative frame for a country you are reading about.
OCCRP	extends	Cross-border investigative reporting with a strong documentary and corporate-records tradition. Where InSight Crime describes the street and territorial structure, OCCRP is more likely to have the company and banking chain.
UNODC	corroborates	Statistical and market analysis on drugs, homicide and trafficking that provides the quantitative counterpart to narrative claims about volumes, routes and prices.
US Treasury sanctions actions	corroborates	Designation packages naming individuals, companies and networks, frequently covering the same actors. A designation converts a journalistic identity into an official one and adds corporate and financial detail.
US Department of Justice	corroborates	Indictments and case documents that are the evidentiary substrate under a large share of the reporting, and the correct citation when a claim needs to be defensible.
International Crisis Group	extends	Field-based political analysis of the conflicts and state weaknesses in which these criminal economies sit, with a stronger focus on negotiation and policy options.
Human Rights Watch	extends	Documentation of abuses by both criminal groups and state forces, which supplies the victim-side and state-complicity dimensions that crime reporting frequently under-covers.
Global Terrorism Database	contradicts	Codes politically framed violence and excludes state action, so a group treated here as criminal may be absent there entirely. The disagreement between the two framings is analytically informative rather than a data error.

Legal, ethical and operational constraints

Three constraints. First, copyright: this is journalism, and reuse beyond quotation and citation needs permission. Second, personal data and the presumption of innocence: the archive names people who are alleged, indicted, convicted, acquitted or simply mentioned, and in most jurisdictions criminal-offence data is a special category attracting elevated protection. Importing a name from an article into an intelligence record strips the legal qualification unless you carry it deliberately, and describing someone as a criminal leader on the basis of a news article can be defamatory, can breach data protection rules and can cause real harm to a person who has not been charged. Carry the status with the name, every time. Third, the duty to sources and victims: the reporting rests on people who took risks to speak, and in several countries in the region has cost journalists their lives. Do not attempt to identify anonymous sources, do not aggregate details in a way that could unmask them, and do not republish identifying details of victims of trafficking, extortion or sexual violence. If your work uncovers a specific person at risk, the route is to the appropriate protection mechanism, not to a publication.

Operational security

Reading a public news site reveals little on its own, but the pattern of your reading can reveal a great deal, and this is a domain where that matters. Sustained, targeted collection against a specific group's coverage or a specific person's name creates a record at your egress point, at any intermediary, and potentially at the site's analytics provider, that says an organisation at your address is researching that subject. If you are working on a matter involving state complicity in a country with capable services, or in an environment where a criminal organisation has access to telecoms or government data, that pattern is a real exposure. Use appropriate infrastructure for sensitive work, avoid corresponding with the newsroom about a live matter unless you have a reason to and your client has agreed, and be aware that contact with journalists covering organised crime is itself sensitive – both for your investigation and, more importantly, for them.

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 InSight Crime is contributing anything, and they are worth baselining now so the answer is available later.

Proportion of claims you rely on that you were able to trace to a document, official action or independent source, tracked over time as your corroboration rate for this source.
Lead time between the source publishing a development and it appearing in official records or in your other collection, measured per country as a test of whether it is genuinely a tipper for you.
Number of entities that had to be split or merged after alias reconciliation, as a direct measure of how much identity noise this source introduces into your graph.
Age distribution of the relationship claims currently active in your network for the region, which reveals how much of your picture is running on stale reporting.
Coverage depth per country in your area of responsibility relative to your own assessment of criminal significance, exposing where the archive's silence is about the newsroom rather than about the country.
Frequency with which a standing profile diff produced an actionable finding, which tells you whether profile monitoring is worth the collection cost for your portfolio.
Share of your client deliverables in which journalism-sourced claims were explicitly labelled as such, as a straightforward quality-control count.

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:

Carry the sourcing basis with the claim from the moment you read it. A claim from an indictment and a claim from an anonymous interview should never look the same three documents downstream, and the only way to guarantee that is to capture the distinction at first contact.
Read chronologically. Criminal structures in this region fragment and rebrand, and only the sequence reveals it; the most recent article describes the present understanding, not the history that produced it.
A brand name is not an organisation. Before you create an actor entity, establish whether the name denotes a coherent structure, a franchise, a label applied by prosecutors, or several unrelated groups using the same word.
Never merge people on an alias alone. Nicknames are the most common identifier in the reporting and the least reliable, and an automated merge on alias will fuse distinct individuals in a way that is very hard to unwind later.
Territorial control is a claim with a date. Map it as such, and refuse to produce a control map that does not show its own vintage on its face.
Trace claims to first publication before counting corroboration. Regional crime reporting recycles heavily, and three sources are frequently one source and two aggregators.
Revenue mix beats org chart. What a group earns from is more stable, more verifiable and more predictive than who reportedly commands it, and it is the attribute that connects the narrative to financial and trade data you can check.
Read the Spanish edition too. Timing and content differ between editions, and for a fast-moving development the local-language version is frequently earlier and more detailed.
Say when your finding is journalism. Clients can weigh a clearly attributed journalistic claim; they cannot weigh a seamless narrative that hides which parts rest on a reporter's interview and which on a court file.

Questions analysts actually ask

Is there an API or a dataset?

No. It is a news website with tagged articles in English and Spanish. Anything structured you get from it, you build yourself by extraction, and the extraction needs analyst judgement to be worth having.

Can I cite it in an intelligence product?

Yes, as investigative journalism, clearly attributed with the article title and date. What you must not do is present its claims as established fact or as your own collection. Stating the provenance makes the product more defensible, not less.

How reliable are the group profiles?

Reliable about the existence, geography and criminal economies of a group; less reliable about internal structure and command, which every narrative source renders more cleanly than reality supports. Check the last update date and cross-check leadership against recent official actions before relying on it.

Can I use the annual homicide figures for cross-country comparison?

Only with the methodological notes attached. The figures come from national institutions with different definitions, counting rules and political pressures, and the compilation says so. Stripping the notes and ranking countries produces a table that is easy to publish and hard to defend.

Three other outlets reported the same thing – is that corroboration?

Frequently not. This outlet is often the origin of the regional narrative, and downstream coverage recycles it. Trace each report to its earliest publication and check whether the later ones did independent work before you count them as separate sources.

Can I republish a profile in a client report?

Not without permission. Treat it as copyrighted journalism, quote briefly with attribution, link to it, and write your own analysis. If you need to reproduce substantial material, ask them.

How do I handle names of people who have not been convicted?

Carry the legal status with the name every time it appears in your holdings and your products – alleged, indicted, convicted, acquitted, mentioned. This is both a data protection requirement in most jurisdictions and a defamation exposure, and stripping the qualifier is the most common way organisations get into trouble with material like this.

Why is there so little on a country I know has serious organised crime?

Because coverage follows grant funding and staffing rather than criminal significance. Read thin coverage as a fact about the newsroom's footprint, and go to national and local sources for that country rather than concluding the problem is small.

How current is a standing profile?

Assume it is not, until you check. Profiles present authoritatively regardless of when they were last revised, and in a landscape where leaderships change through captures and extraditions within months, an unverified profile is one of the easier ways to put a stale claim into a live assessment.

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:

Unstructured journalism, so everything structured is your construction; document your extraction rules so that two analysts working the same archive produce comparable entities.
Person and organisation entities extracted here should reconcile to sanctions and designation identifiers where possible, since those provide the only authoritative identity anchors in this domain.
Relationship claims map to STIX 2.1 relationship objects only if you carry the source reference and confidence explicitly; an edge without provenance is worse than no edge for this material.
Criminal economy categories align loosely with international classifications of organised criminal markets, which is the frame to reconcile them to rather than inventing a local taxonomy.
Geographic references are frequently at municipality or neighbourhood level and should be geocoded against national administrative gazetteers rather than a global coarse-resolution one.
Article-level citation with title, outlet, date and URL is the minimum standard for any downstream use, and should survive into exported STIX or CSV as a note or external reference.
Legal status qualifiers – alleged, charged, indicted, convicted – are part of the data, not editorial decoration, and any schema you build must have a field for them.

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.

InSight Crime — InSight Crime. The source itself: investigations, country profiles, criminal group profiles and the annual homicide compilation, in English and Spanish.
Global Initiative Against Transnational Organized Crime — GI-TOC. Global-scope analysis and comparative indexing of organised crime and state resilience, providing the frame in which a regional finding can be placed.
OCCRP — Organized Crime and Corruption Reporting Project. Cross-border investigative journalism with strong corporate and financial documentation, complementing territorial and street-level reporting.
UNODC — United Nations Office on Drugs and Crime. Drug market, homicide and trafficking statistics and analysis – the quantitative check on narrative claims about volumes and routes.
US Department of the Treasury — US Treasury. Sanctions designation packages naming networks, individuals and companies, frequently the first official corroboration of a reported structure.
US Department of Justice — US DOJ. Indictments and case materials that underpin much of the reporting and provide the citable, evidentiary version of a claim.
International Crisis Group — Crisis Group. Field-based conflict analysis covering the political context in which these criminal economies operate.
Human Rights Watch — HRW. Documentation of abuses by criminal groups and by state forces, supplying the victim and state-complicity dimensions.
Committee to Protect Journalists — CPJ. Context on the risks faced by reporters covering organised crime in the region, which is the reason certain networks are systematically under-reported.

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 ingests the reporting as dated, attributed documents, promotes entities only under analyst review, and models every alliance and control claim as a sourced assertion with a decay clock rather than as a fact in the graph.. Browse the full source catalogue, or follow any tag above into the rest of the library.