INTERPOL Stolen Works of Art: Intelligence Source Guide

INTERPOL's Stolen Works of Art database holds descriptions and images of almost 57,000 objects, entered only by national central bureaux and a small set of partner organisations. It is the only international register carrying certified police information on stolen cultural property, and since the ID-Art app launched it is searchable by anyone with a phone.

At a glance

Source	INTERPOL Stolen Works of Art
Category	Conflict, Crime & Human Security › Cultural Property & Counterfeiting
Homepage	https://www.interpol.int/Crimes/Cultural-heritage-crime
Machine interface	`https://www.interpol.int/en/How-we-work/Databases`
Format	`HTML`
Access	Open — no account required
Disciplines	Legal Intelligence, Open Source Intelligence
Mission domains	Art & Antiquities Trafficking

International stolen cultural-property database. — as catalogued in the platform’s own source registry.

The database is operated by INTERPOL's cultural heritage crime unit and has been running since the 1990s. Entries are created from information supplied by member countries through their national central bureaux, and by a defined set of authorised international partners including UNESCO, ICOM and ICCROM; no other route into the database exists. INTERPOL states that only fully identifiable objects are entered, which in practice means a description sufficient to distinguish the item and, in almost all cases, a photograph. Access has widened in stages. Authorised users, including art market professionals, museums, dealers and researchers, may apply for an account to search the database directly in real time, using conventional descriptive fields or by uploading an image for comparison against the holdings through image-matching software. Since the release of the ID-Art mobile application, the same holdings are searchable by the general public without any login, alongside three further functions: creating a private inventory of a collection in Object ID format, reporting an item as stolen where a search produces a match, and recording cultural sites at risk with location, description and images. INTERPOL states it cannot read inventories or site cards created by users unless they are reported.

What distinguishes this source from every commercial register is provenance of the record itself. An entry exists because a state's police service reported a theft through its national central bureau, which means there is a criminal complaint, a reporting jurisdiction and an authority that can be contacted. That is why the database is named in the 1995 UNIDROIT Convention as one of the international registers relevant to the concept of due diligence, and why a search timestamp from the ID-Art application can be retained as evidence that a due diligence check was performed. For an investigator this makes it the first register to check rather than the last: it costs nothing, it can be searched by image, and a hit gives you a route into a live criminal case rather than a private commercial negotiation. It is also the register that connects an object to a person, since a theft report exists within an investigation. The trade-off is that its holdings are far smaller than the commercial registers, because the threshold for entry is a police report from a state willing to make one, and that threshold excludes most of what actually moves.

Who publishes it, and why that matters

INTERPOL is an intergovernmental organisation of member countries, and it does not investigate; it facilitates cooperation between national police services. The database is therefore a shared repository whose content is entirely dependent on member states choosing to report and on the quality of what they submit. Strict data processing rules govern what may be entered and by whom, which is what makes the holdings small and reliable rather than large and uneven, and it is the deliberate design choice that defines the source. The organisation is funded by member country contributions and by external partners; the ID-Art application was developed with support from the INTERPOL Foundation for a Safer World and is published free in the organisation's four official languages, Arabic, English, French and Spanish. Continuity is not a practical concern. The operational concern is different: because entry is mediated by national central bureaux, both the speed and the completeness of the record vary enormously with a country's institutional capacity and its willingness to publicise a loss, and there is no mechanism by which anyone outside that chain can add a record.

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
`object_category`	enum	The class of cultural object: painting, sculpture, archaeological artefact, religious item, manuscript, coin, furniture and similar. Category is also a search facet in both the web interface and the mobile application.	Specialist expertise; auction department; ICOM Red List category for the same material.
`artist_or_maker`	string	Attributed artist or maker where known. Present for fine art and largely absent for archaeological and ethnographic material, which is precisely the material most likely to be trafficked.	Catalogue raisonne; auction and museum records; art historical literature.
`title`	string	The object's title or common name as reported. Titles vary between languages, catalogues and periods, so exact-match searching on title alone will miss records that are plainly the same object.	Sale catalogues; exhibition histories; press reporting of the theft.
`medium_and_technique`	string	Material and method of manufacture. Searchable as a field and, for archaeological material, often the most discriminating text available in the absence of an artist name.	Scientific analysis; comparison with excavated parallels; conservation records.
`dimensions`	string	Measurements as reported by the notifying country. The most reliable non-image discriminator, and also the field most affected by inconsistent conventions between reporting services.	Physical examination of an offered object; catalogue entry comparison.
`image`	array	Photographs of the object. INTERPOL states only fully identifiable objects are entered, and the image is what makes the visual search function work; an object reported without one is very much weaker in the system.	Image-matching search; comparison with listing and catalogue photographs.
`country_of_theft`	string	The reporting jurisdiction. It identifies the national central bureau and the police service that hold the underlying case, which is the operationally important consequence of a match.	National central bureau contact; the reporting force; export licensing regime of that state.
`date_of_theft`	timestamp	When the object was reported taken. For institutional thefts this is usually precise; for losses discovered during an inventory it may be a broad window, and for wartime losses it can be a period rather than a date.	Case chronology; market appearance timeline; limitation periods in the relevant jurisdiction.
`reference_number`	string	The INTERPOL record identifier for the object. Cite it in correspondence with any authority, since it is the key that lets a national central bureau retrieve the underlying report.	National central bureau enquiry; case file retrieval.
`notifying_entity`	enum	Which authorised entity supplied the record: a national central bureau or one of the permitted international partners. Only these routes exist, which is the source of both the database's reliability and its limited size.	Authority to contact; assessment of report completeness.
`search_timestamp`	timestamp	Generated by the ID-Art application when a manual search is saved. INTERPOL states these timestamps may be used as proof that due diligence checks were carried out, in line with the UNIDROIT Convention.	Provenance file; transaction record; evidence of good faith.
`site_card`	array	Records created in the application to document a cultural site at risk, with geographic location, description and images in Object ID format, for use as evidence or as a basis for reconstruction after looting or destruction.	Heritage protection planning; damage assessment; post-conflict documentation.

Coverage — and what is not in it

The database holds close to fifty-seven thousand objects, which is small beside commercial registers and is small by design: an entry requires a police report routed through a national central bureau, plus a description sufficient to identify the object and, effectively, a photograph. Coverage therefore tracks two things. The first is state reporting behaviour: countries with specialist cultural property police units and a habit of international cooperation report far more than countries without, and some states report almost nothing regardless of how much is stolen from them. The second is institutional versus private loss: thefts from museums, churches and public collections are more likely to be reported and better documented than thefts from private houses, and both are far better represented than looting from archaeological sites, which produces objects nobody can report because nobody recorded them. Coverage is global in principle with no geographic restriction. Updates are continuous as member countries submit, and objects are removed or updated when recovered, so the holdings turn over rather than only accumulating. There is no published historical archive of removed entries.

Known blind spots

Absence of evidence here is not evidence of absence. These are the conditions under which INTERPOL Stolen Works of Art will not show you something that is nevertheless real:

An object must have been documented before it was taken. Illicitly excavated antiquities and undocumented private holdings cannot be reported stolen because no record of their existence preceded the theft, and that describes the majority of trafficked archaeological material.
A state that does not report produces no entries. Reporting depends on national police capacity, on a functioning national central bureau, and on a government willing to publicise the loss, and several countries with severe looting problems appear very lightly in the holdings.
Only fully identifiable objects are entered, which excludes generic, mass-produced and poorly documented items even when they were genuinely stolen, and excludes objects whose owners never photographed them.
Thefts during conflict frequently go unreported for years or permanently, because the institutions that would report them have ceased to function, and the objects reach the market long before any record exists.
The database records objects, not networks. It will not tell you who handled an item, where it moved, or that several entries share a common thread, and any pattern analysis has to be done outside it.
Wartime and colonial-era losses are largely outside its scope. Claims from the 1933-45 period and from colonial acquisition are handled through specialist registers and restitution processes, not through a police theft database.
Image search depends on the quality of both images. A poor archival photograph of the object as it was, matched against a poor listing photograph of the object as it now appears after cleaning, restoration or reframing, will frequently fail.
Recovered objects are removed or updated, so a search today cannot tell you what was on the database last year, and a negative result gives you no way to establish whether an entry once existed.
Access for full-featured searching still requires an application and approval, so an authorised-user account is not instantaneous, and the public mobile application, while genuinely open, exposes the holdings through a narrower interface.

Write the blind spot into the product. A statement that something “was not observed in INTERPOL Stolen Works of Art” 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

Two routes, both free. The ID-Art mobile application is published for Apple and Android devices, requires no username or password to search the database, and offers manual search on descriptive fields as well as visual search by photographing an object. It is available in Arabic, English, French and Spanish. The second route is an authorised-user account for the web database, obtained by completing INTERPOL's application form, which is open in principle to anyone able to justify access; art market professionals, museums, insurers, researchers and law enforcement all hold such accounts. The web interface supports the same descriptive and image-based searching with a fuller working environment. For law enforcement the operative route is neither of these: it is the national central bureau, which can query, submit and act on records within the police channel and is the correct interlocutor for anything that will become a case. Note that the database cannot be added to except through a national central bureau or an authorised partner organisation, so reporting a theft means reporting it to the police first.

Licence

This is an intergovernmental police information system, not an open data product, and the terms should be read that way. Search results are provided for the purpose of checking whether an object is recorded as stolen, and INTERPOL's data processing rules govern what may be done with information obtained from its systems. Bulk extraction, republication of holdings, or building a derived commercial database from search results is outside that purpose and should not be attempted. Conversely, the intended uses are broad and encouraged: due diligence before a purchase or a loan, checking objects encountered in the course of business, and retaining a dated search record as evidence of the check. The application's terms of use and privacy policy are published alongside it and are short enough to read before deployment in an organisation. If your intended use extends beyond checking specific objects, ask through the appropriate channel rather than assuming.

Rate limits and fair use

No published quota applies to ordinary use, and none is needed for the way the system is meant to be used, which is object by object. The service is not designed for programmatic bulk querying and there is no public API; attempting to automate against it would be both technically fragile and contrary to the purpose for which access is granted. For an organisation checking large volumes, such as an auction house catalogue, the appropriate approach is an authorised-user account and a documented internal process, and where volumes are genuinely large, a conversation with INTERPOL or the relevant national unit about how to handle it. The practical etiquette point is that this is a police system supported by member country contributions: use it for real enquiries, not for exploration.

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 INTERPOL Stolen Works of Art 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
ID-Art visual search	HTML	per object	Photograph the object and let image-matching compare it against the holdings. The fastest first check available anywhere and the right opening move for any physical object in front of you.
ID-Art manual search	HTML	per object	Search by object type, medium, technique, title, artist or country of provenance, and save the search so the timestamp is retained as a due diligence record.
Authorised-user web search	HTML	per object; continuous during a catalogue review	The fuller interface, obtained by application. Appropriate for professional users handling volume, and the route that produces a workable audit trail inside an organisation.
National central bureau enquiry	HTML	per case	The law enforcement channel. The only route by which a record can be created or a match acted on within a criminal investigation, and the correct destination for any confirmed hit.
Object ID inventory and site cards	JSON	on documentation	Create structured records of collections and at-risk sites in the application. This is preventive collection: it is what makes a future theft reportable at all.

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 the source as a manual check with a defined route — In sources.php, record the database as a per-object check with two access routes and no API, so nobody in the workflow builds an expectation of automated matching that the source cannot support.
Model the object with Object ID fields — Create the object in entity.php using the Object ID field set: type, materials and techniques, measurements, inscriptions and markings, distinguishing features, title, subject, date, maker, and a written description with photographs. Searching and reporting both depend on this being complete.
Log every search as a dated event — Record each check through import.php as an event against the object, carrying the date, the route used and the result. The date is the evidentially significant part, and storing a result without it destroys the due diligence value.
Attach the reference number on a match — Where a hit occurs, store the INTERPOL reference and the reporting country against the object, since those two fields are what let a national central bureau retrieve the underlying case.
Cross-run the parallel registers — Trigger checks against the commercial register, national police databases and the wartime-losses registries in the same case workflow, and record each result separately so the file shows the full set of checks performed.
Bind the referral pathway to the country — Link the country of theft to the corresponding national central bureau and specialist unit in le-contacts.php and agencies.php, so a match immediately produces a named destination rather than a research task.
Restrict distribution of match records — A hit concerns a live criminal matter and identifiable parties. Keep it inside the case with access controls in cases.php, and exclude it from automated exports and shared reporting until the responsible authority has been engaged.
Schedule re-checks for held objects — Configure alerts.php so any object held on consignment, under consideration or in storage is re-searched periodically, because entries are added continuously and a check performed six months ago no longer reflects the holdings.

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.

Judged by precision, this is the best cultural property register available: every record originates from a police report routed through a national central bureau or an authorised international partner, only fully identifiable objects are accepted, and images are effectively required. When it returns a match, the match means something specific and actionable, and there is an authority behind it. Judged by recall it is weak, and honestly so: fifty-seven thousand objects is a small fraction of what has been stolen, and the gap is largest in exactly the categories and countries where trafficking is most active. The image-matching capability is a genuine advance and materially improves the odds on objects that have been photographed well at both ends, while performing poorly where the archival photograph is bad or where the object has been altered. The soundest way to judge a result is by understanding the reporting chain behind it: a match tells you a state reported this object, and a negative tells you nothing at all about whether the object was stolen, only that no state reported it in terms that this system could record.

Characteristic false positives

A negative result is treated as clearance. The database holds only what states reported, so an absence means no report exists in a system with a high entry threshold, not that the object has a clean history. For archaeological material a negative result is close to uninformative.
Image matching produces plausible near-misses. Objects of a common type, particularly repeated religious statuary, furniture and coins, generate visual matches that are the same class rather than the same object, and a confident identification from a screen without physical comparison is unsafe.
Image matching fails on the same object. Cleaning, restoration, reframing, relining and simple differences in lighting and angle between an old archival photograph and a modern listing image defeat comparison, so a genuine stolen object can return nothing.
The reported description does not match the market description. A theft report written by an officer at a scene and a catalogue entry written by a specialist can describe the same object in incompatible vocabulary, particularly across languages, so text search misses records that a human would recognise instantly.
A match is read as proof of who stole the object. It establishes that the object was reported stolen from somewhere, not that the current holder knew, participated or acquired it in bad faith, and conflating the two is both an investigative and a legal error.
Recovered objects have been removed, so an old note that an object was on the database cannot be verified later, and a case built on a remembered hit rather than a saved record has no evidence behind it.
Objects reported by different countries under different conventions produce duplicate or near-duplicate entries, and treating two records as two objects inflates the picture where a cross-border theft was reported at both ends.
The date of theft is the date of report or discovery rather than the date of the crime, particularly where a loss was found during an inventory, which distorts any timeline that treats it as the moment the object entered the market.

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

Records here behave differently from those in a static reference. Entries are added continuously as states report, and are removed or amended when objects are recovered, so the holdings turn over. That gives a search result a short useful life in both directions: a negative result decays because new reports arrive, and a positive result can become obsolete because the object was recovered and the entry closed. Either way, the practical rule is that a search is a statement about a moment, which is precisely why the application timestamps saved searches and why those timestamps have evidential value. There is a second, slower kind of ageing in the record content: object descriptions and photographs are captured at the time of loss, and an object that reappears years later after cleaning, restoration or a change of frame or mount may no longer resemble its own record. A stale record looks like an outstanding entry for an object recovered quietly through a private settlement, or a search certificate for an object that was reported stolen a week after it was checked.

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

Art & Antiquities Trafficking

Yields these data points

How each sector uses INTERPOL Stolen Works of Art

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

Relevant to cultural property protection obligations in and after conflict. The site card function in the application is the operationally distinctive feature: personnel can document the condition of monuments, archaeological and underwater sites with location, description and images, producing structured records usable as evidence of damage or as a basis for reconstruction. For objects encountered during operations, the search function gives an immediate check, but the correct handling is always through the legal adviser and the cultural property protection channel rather than any independent action. Bear in mind the structural gap that matters most in theatre: material looted from unexcavated sites cannot be in this database, so a negative result on an antiquity recovered in a conflict zone carries no implication of legitimacy.

🕵 National intelligence

The database is a narrow but high-quality input into work on trafficking networks and conflict financing. Its analytical value lies less in the objects than in the reporting behaviour around them: which states report, how quickly, and what they choose to describe tells you something about institutional capacity and political will that is relevant well beyond cultural property. A match provides an entry point into a criminal case in a named jurisdiction, which is often more useful than the object itself. For network analysis the database is a poor instrument, since it records objects without relationships, and the pattern work has to be done against freight, corporate, sanctions and financial data with the object as a single node.

👮 Law enforcement

This is the primary international check and, for police services, the primary reporting route. Practically: report through your national central bureau so an object enters the international system, cite the reference number in any international correspondence, and use the image search on objects encountered in searches, seizures and market monitoring. The application's saved-search timestamps matter in a second way for enforcement, since they can establish whether a dealer or purchaser performed a check and when. A match is intelligence pointing to an existing case in another jurisdiction, and the next step is the national central bureau rather than direct contact with a foreign force. Where the object is in the hands of a purchaser, remember that the register speaks to the object's status and says nothing about that person's knowledge.

🔍 Private investigation and corporate security

The first check in any art or antiquities engagement, because it is free, fast and image-capable. Do it before commissioning paid searches, since a hit here resolves the matter and costs nothing, and record the saved-search timestamp in the client's provenance file, since INTERPOL states these may serve as proof that due diligence was carried out. Advise clients honestly about what a clean result covers: a small, high-quality register of objects that states reported stolen, which for antiquities is a very partial picture. An authorised-user account is worth obtaining if art work is a recurring part of the practice, because the web interface handles volume better than a phone.

📰 Journalism and OSINT media

Uniquely useful among cultural property sources because a journalist can actually search it: the application requires no account and supports image search, so an object seen in a catalogue, a fair or a social media listing can be checked directly. Two cautions apply. A match means the object was reported stolen, not that the current holder is a criminal, and reporting must keep that distinction sharp or it becomes defamatory. And a negative result is not a story about the object being clean; it is usually a story about how little of the trade is reportable at all, which is the more important piece anyway. Where a live case exists, coordinate with law enforcement before publication, because publicity can move an object out of reach.

🌍 NGO, humanitarian and human rights

The most valuable engagement is documentation rather than searching. The application's inventory function records objects in Object ID format, and the site card function records at-risk cultural sites with location and images, and both directly address the structural failure that makes recovery impossible: undocumented objects cannot be reported stolen. Supporting museums, religious institutions and archaeological projects in source countries to inventory their holdings is high-leverage, low-cost protection work. INTERPOL states that user-created inventories and site cards are not readable by the organisation unless reported, which is a relevant assurance when working with communities cautious about sharing information about their own heritage with police systems.

🎓 University and research

A research resource and a research object. As a resource it supports work on the visible stolen art market, on recovery rates and on the relationship between reporting behaviour and institutional capacity. As an object of study it illustrates how an intergovernmental body balances data quality against coverage, and what happens when a police register is opened to the public through a consumer application. Researchers should be careful with quantitative claims: the holdings reflect state reporting practice rather than theft, so any rate computed against them measures reporting as much as crime. Access for research purposes runs through the same authorised-user application as for anyone else, and bulk extraction is neither available nor appropriate.

Playbook: working INTERPOL Stolen Works of Art 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 — Start here, before anything paid

The image search is free, immediate and covers police-certified records. Any art or antiquities matter should begin with this check, because a hit resolves the question at zero cost and a miss tells you what the rest of the work has to establish. Analysts who reach for the commercial register first are spending money to answer a narrower question.

Phase 2 — Document the object to Object ID before searching

Type, materials and techniques, measurements, inscriptions and markings, distinguishing features, title, subject, date, maker, description and photographs. The application uses this structure and so does every other register, so building it once serves every subsequent check and, if the object is later stolen, makes it reportable.

Phase 3 — Search visually and textually, and treat them as different tests

Photograph the object for image matching, then search independently on category, medium, dimensions and any maker attribution. The two methods fail in different ways: image matching fails on altered or badly photographed objects, text search fails on vocabulary and language differences. Doing only one leaves a predictable gap.

Phase 4 — Save the search and preserve the timestamp

INTERPOL states saved-search timestamps may be used as proof that due diligence was carried out in line with the UNIDROIT Convention. Save the search, export or record it into the case file, and note the date alongside the transaction timeline. This is the cheapest evidential step available in the whole cultural property workflow.

Phase 5 — Interpret a negative result honestly and in writing

Write down what a clean result actually covers: a register of objects that states reported stolen, with a high entry threshold, holding tens of thousands of items. For archaeological material state plainly that a negative result is expected and uninformative. If you do not write this, someone will later read your clean result as a clearance.

Phase 6 — Escalate a positive match through the police channel

A hit points to a criminal case in a named jurisdiction. Route it through your national central bureau, cite the record reference, and resist the urge to contact the reporting force, the claimant or the current holder directly. Independent contact can compromise an existing investigation and can put the object beyond reach.

Phase 7 — Separate the object's status from anyone's culpability

The record establishes that the object was reported stolen. It says nothing about how the current possessor acquired it or what they knew. Keep those questions distinct in every note and every communication, because merging them is both an analytical error and a defamation exposure.

Phase 8 — Run the parallel registers as a documented set

Check the commercial register, the relevant national police databases, the wartime-losses registries and the category typologies, and record each result with its date. A file that shows a set of checks is defensible; one that shows a single check invites the question of what else you did not look at.

Phase 9 — Apply the typologies where the register cannot help

For undocumented archaeological and ethnographic material, the ICOM Red List categories do the work this database cannot, converting an unmatched object into a documented provenance requirement. Treat the two sources as a pair rather than as alternatives.

Phase 10 — Use the site and inventory functions preventively

Where you have access to collections or sites at risk, create structured records now. The single greatest constraint on recovery is that most stolen cultural property was never documented, and this is the only widely available free tool that closes that gap at the point where it matters.

Phase 11 — Re-check before completion and periodically while holding

Entries arrive continuously, so a check performed at the start of a consignment or a loan is not the check that matters at the end of it. Schedule a re-search before any transaction completes and periodically for objects held, and record each one.

Phase 12 — Write the referral pathway into the product

Whatever you deliver should name the authority to be contacted for the country of theft, the record reference where one exists, and what has already been reported. An identification that does not end in a named recipient is an observation rather than an intervention.

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
ICOM Red Lists	extends	Category typologies for undocumented material that cannot appear in any stolen-object register. The necessary complement for archaeological and ethnographic items, where a negative result here is expected.
Art Loss Register	extends	The much larger commercial register, covering insurance-driven registrations, title disputes and liens that never reach a police report. Different population, chargeable, and the market's standard due diligence route.
Object ID	prerequisite	The documentation standard the ID-Art application uses for inventories and site cards, and the format in which an object must be described to be reportable at all.
Carabinieri Command for the Protection of Cultural Heritage	corroborates	Italy's specialist unit maintains one of the largest national stolen art databases and a body of operational practice that goes well beyond what the international register holds.
Lost Art Database	extends	Wartime and Nazi-era losses, which fall outside a police theft register but frequently arise in the same provenance enquiries.
Central Registry of Information on Looted Cultural Property 1933-1945	extends	The principal public resource for 1933-45 losses, restitution claims and the associated case law.
UNIDROIT 1995 Convention	prerequisite	The instrument that gives a documented register check legal consequence, and which names international registers of this kind in the due diligence context.
World Customs Organization	corroborates	The border enforcement framework through which flagged objects are actually intercepted, and the source of joint operational activity against cultural property trafficking.
UNESCO fight against illicit trafficking	prerequisite	The convention framework, states parties and national implementation resources that determine what a match means legally in a given country.

Legal, ethical and operational constraints

Information obtained from an international police information system carries handling obligations even when the interface is a consumer application. Treat a match as sensitive: it concerns a live criminal matter, identifiable victims and a current possessor who may be entirely innocent. Publishing or circulating an assertion that a named person or business holds stolen property before any adjudication is actionable in most jurisdictions, and the record itself does not support that assertion, only that the object was reported stolen. In several countries handling stolen cultural property is a criminal offence regardless of knowledge, and in others knowledge or recklessness is the test, so advice on the possessor's position is jurisdiction-specific and should come from a lawyer rather than from an analyst. Where an object is potentially evidence, custody and continuity rules apply from the moment you form that view, and the correct step is a report to the competent authority rather than further independent enquiry. Finally, the register is one input into a due diligence standard set by conventions and national law; treating a clean search as legal clearance misstates what the check does.

Operational security

Searching through the mobile application requires no account, which is a genuine privacy feature and unusual for a police system, but the application is still a network client and its terms and privacy policy should be read before organisational deployment. The exposure that matters is downstream and human. Enquiries about a specific object directed at a dealer, an auction house or a national authority signal interest and can cause an object to be withdrawn, moved or destroyed, and in the antiquities trade that happens quickly. Where a matter is live, coordinate with law enforcement before any approach. For organisations documenting collections or sites in the application, note INTERPOL's statement that it cannot read user-created inventories or site cards unless they are reported as stolen; that assurance is worth having in writing when working with communities or institutions that are cautious about sharing heritage information with police systems, and worth verifying against the current terms of use.

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 INTERPOL Stolen Works of Art is contributing anything, and they are worth baselining now so the answer is available later.

Share of art and antiquities matters in which this register was checked before any paid search, which is the direct measure of whether your workflow uses the free, police-certified source first.
Proportion of checks performed both visually and textually, since the two methods fail differently and doing only one leaves a predictable gap.
Share of searches with a preserved timestamp filed in the case record, which is what converts a check into evidence of due diligence.
Median interval between the last check and transaction completion, and the proportion of transactions with a re-check immediately before completion.
Number of matches escalated to a national central bureau within a defined time of identification, tracked because delay is what allows a flagged object to move.
Count of objects or sites documented to Object ID standard through your work, which is the only metric that improves the register's future coverage rather than exploiting its current coverage.
Proportion of your reports that state explicitly what a negative result covers, particularly for archaeological material where the negative is expected.

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:

Check this register first. It is free, image-capable and police-certified, and a hit resolves the question before any money is spent elsewhere.
Save the search and keep the timestamp. INTERPOL states these timestamps can evidence due diligence under the UNIDROIT framework, and an unsaved search has no evidential existence at all.
For antiquities, expect nothing and say so. Objects looted from unrecorded contexts cannot be here, so a clean result is the expected outcome and carries no information about legitimacy.
Run image search and text search as separate tests. They fail on different objects, and treating either as a complete check produces a predictable class of misses.
A match is about the object, not about the holder. Keep the object's reported status and any person's culpability in separate sentences from the first note onwards, because merging them creates legal exposure and misdirects the investigation.
Route hits through the national central bureau. Direct contact with a foreign force, a claimant or a possessor can compromise a live case and can put an object beyond recovery.
Descriptions were written at the moment of loss. Titles, attributions and dimensions recorded then may not match a modern catalogue entry for the same object, so search generously and read results as a specialist would.
Documenting objects and sites is the highest-value use of this system. Recovery is impossible for undocumented material, and the inventory and site card functions are the only free, widely available way to fix that at the source.
Re-check before completion. Entries arrive continuously, and a due diligence file that ends with a search from six months earlier has a gap exactly where the risk sits.

Questions analysts actually ask

Can anyone search the database?

Yes, through the ID-Art mobile application, which requires no login and supports both manual and image-based searching in Arabic, English, French and Spanish. A fuller web interface is available to authorised users, who apply through INTERPOL's application form; art market professionals, museums, insurers and researchers hold such accounts alongside law enforcement.

How large is it, and why is it so much smaller than commercial registers?

It holds almost fifty-seven thousand objects. It is smaller because entry requires a police report routed through a national central bureau or an authorised partner such as UNESCO, ICOM or ICCROM, and because only fully identifiable objects are accepted. That threshold is what makes a match meaningful, and it is also what keeps the holdings small.

How do I get an object added?

You cannot add it yourself. Report the theft to the police in the jurisdiction where it occurred, and ask that it be transmitted to INTERPOL through the national central bureau. Supply an Object ID standard description with photographs and dimensions, because incomplete reports do not meet the entry threshold. The ID-Art application can export an inventory record to support that report.

Does a clean result mean the object is safe to buy?

No. It means no member state has reported that object as stolen in terms this system could record. For fine art with a documented history that is meaningful; for archaeological or ethnographic material it is close to uninformative, because such objects are generally not documented before they are taken. Provenance documentation, not a register check, is what answers the question for that material.

How reliable is the image matching?

It is a genuine advance and works well when both the archival photograph and your image are good and the object is unaltered. It fails in two directions: near-misses on common object types that share a form, and misses on objects that have been cleaned, restored, reframed or photographed under very different conditions. Use it alongside text search, never instead of it.

What should I do if I get a match?

Do not contact the reporting country, the claimant or the current holder directly. Record the reference number, preserve the search, and report through your national central bureau or the relevant national police unit. A match points to a live criminal case in another jurisdiction, and independent contact can compromise it.

What is the site card function for?

It records the condition of cultural sites at risk, including historical monuments and archaeological and underwater sites, with geographic location, description and images. The resulting Object ID record can serve as evidence or as a basis for reconstruction if a site is later looted, destroyed or vandalised, and users can report illicit excavations to national authorities.

Does INTERPOL see the inventory I create in the app?

INTERPOL states that it cannot read Object ID records or site cards created by users, and sees only those reported as stolen within the application. That assurance matters when working with private collectors or with institutions in sensitive contexts, and it is worth confirming against the current terms of use before organisational deployment.

Is the search itself legally useful?

Yes. The database is named in the 1995 UNIDROIT Convention among the international registers relevant to due diligence, and INTERPOL states that timestamps on saved searches in the application may be used as proof that the necessary checks were performed. Saving the search and filing it with the transaction record is therefore worth the few seconds it takes.

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:

Object ID is the documentation standard used for the application's inventories and site cards, and the format in which cultural objects should be described for reporting to any register.
The 1995 UNIDROIT Convention on Stolen or Illegally Exported Cultural Objects names international registers of this kind in relation to due diligence, which is what gives a dated search record legal weight.
The 1970 UNESCO Convention and its national implementations govern export, import and transfer of ownership, and determine what a country of theft can require on recovery.
The 1954 Hague Convention and its protocols govern protection of cultural property in armed conflict and underpin the site documentation use case.
INTERPOL's own data processing rules define who may enter records and how information obtained from its systems may be handled, and they are the reason the holdings are small and reliable.
National central bureau procedures are the operational standard for creating, querying and acting on records, and are the correct channel for any matter that will become a case.
Platform exports carry object entities, case records and referral contacts in STIX 2.1, MISP, CSV, JSON and JSONL, so a cultural property matter shares structure with the rest of a caseload.

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.

Stolen Works of Art Database — INTERPOL. The authoritative description of the database, its size, who may enter records and how to apply for access. The single page to read before using the source.
ID-Art mobile app — INTERPOL. What the public application does: database search by description and image, private inventories in Object ID format, reporting a stolen item, and recording sites at risk.
Cultural heritage crime — INTERPOL. The programme overview, covering how the organisation works on cultural property crime, its partnerships and the guidance for institutions and the public.
INTERPOL databases — INTERPOL. Context for how the works of art database sits alongside the organisation's other systems and how access through national central bureaux operates.
Object ID — International Council of Museums. The documentation standard behind the application's inventory function, and the prerequisite for any object being reportable at all.
ICOM Red Lists — International Council of Museums. The category typologies that cover the undocumented material this database structurally cannot hold. The necessary companion source for antiquities.
UNIDROIT Convention on Stolen or Illegally Exported Cultural Objects — UNIDROIT. The convention text and status list, including the due diligence provisions that give a documented register check its legal effect.
UNESCO fight against illicit trafficking — UNESCO. Convention framework, states parties, national implementation and the cooperation mechanisms through which restitution actually happens.
World Customs Organization — WCO. Customs enforcement practice and joint operations, which is where objects flagged against this database are most often physically intercepted.
Carabinieri Command for the Protection of Cultural Heritage — Arma dei Carabinieri. The most developed specialist cultural property police service, whose national database and operational model illustrate what systematic national reporting produces.

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 documents each object to Object ID standard, records every register check as a dated event with its result preserved, binds a match to the reporting country's national central bureau as a named referral, and re-checks held objects before any transaction completes.. Browse the full source catalogue, or follow any tag above into the rest of the library.