Cross-Border Criminal Justice Cooperation: What Europol, Eurojust, and the European Arrest Warrant Actually Do

Three Institutions, Three Jobs: Europol, Eurojust, and the European Public Prosecutor's Office

Europol, Eurojust, and the European Public Prosecutor's Office sound like interchangeable pieces of the same EU-level police apparatus. They are not, and the clearest starting point is what Europol cannot do. Europol has no power of arrest, no independent investigative authority, and cannot operate on member-state territory without national law enforcement carrying out the operation itself: every arrest made in an Europol-supported case is executed by national police under national law. Europol, formally the European Union Agency for Law Enforcement Cooperation, operates under Regulation (EU) 2016/794, whose statutory mission is to support and strengthen cooperation among the competent authorities of the member states in preventing and combating serious cross-border crime, terrorism, and crime affecting a common EU policy interest. Its operational backbone is SIENA, the Secure Information Exchange Network Application, which in 2024 carried more than 2 million messages, its highest volume on record, alongside an 18.6 percent increase in connected competent authorities. Operational work is organized into standing Analysis Projects, among them AP Cyborg for cyber-dependent crime, AP Hydra for terrorism, and AP Furtum for property crime.

Regulation (EU) 2022/991 expanded Europol's mandate in 2022 to include direct cooperation with private parties, processing of large datasets for ongoing investigations, and a research and innovation role touching AI-related tooling. That expansion remains contested rather than settled: the European Data Protection Supervisor formally objected and initiated legal action, arguing the amendment retroactively authorized data processing that had already occurred outside the prior legal framework, and that data belonging to people with no established link to any crime would be processed on the same footing as crime-linked data.

Eurojust, formally the European Union Agency for Criminal Justice Cooperation, operates under Regulation (EU) 2018/1727 and supports national judicial authorities, prosecutors and investigating judges, rather than police. It hosts three secretariats: the European Judicial Network, the Joint Investigation Teams Network, and the Genocide Network. Its 2024 caseload corrects any assumption of ever-rising volume: 981 European Arrest Warrant-related cases, roughly 20 percent lower than 2023. The two agencies' sharpest shared operational tool is the Joint Investigation Team, a temporary, case-specific team of prosecutors, police, and judges from two or more states, typically running 12 to 24 months. No current, precise count of active Joint Investigation Teams for 2025 or 2026 is available from the sources behind this article. A companion piece in this workspace, covering cybercrime legislation across EU jurisdictions, treats Europol and Eurojust's cybercrime-specific operational record, including Operation Endgame and Eurojust's own 2018 cybercrime-caseload figures, in depth, while this article's concern is the two agencies' general institutional mandate.

National coordination bodies sit beneath this EU layer: Germany's Bundeskriminalamt and GTAZ, its joint counter-terrorism coordination center, handle cross-border coordination nationally, a subject for a separate discussion of police oversight rather than this article.

The European Public Prosecutor's Office (EPPO) is the one genuinely supranational body in this trio, established under Regulation (EU) 2017/1939 and operational since 2021. Unlike Europol and Eurojust, it prosecutes directly, over crimes against the EU budget. By the end of 2025 it was investigating 3,602 active cases, up 35 percent on 2024, with an estimated €67.27 billion in damage and a conviction rate near 95 percent. Its 23rd participating state, Hungary, was approved on 10 July 2026, leaving only Denmark and Ireland outside by opt-out, the same pattern that recurs across the other instruments covered here.

The European Arrest Warrant: Mutual Recognition Applied to People

Framework Decision 2002/584/JHA replaced classical, diplomatically mediated extradition among EU member states with direct court-to-court surrender, built on mutual recognition: a judicial decision issued in one member state is executed in another as if it were the executing court's own decision. A warrant may be issued for prosecution purposes where the underlying offense carries a maximum custodial sentence of at least 12 months in the issuing state, or for execution of an already-imposed sentence where at least 4 months remain to be served.

The Framework Decision's most significant substantive innovation is Article 2(2)'s abolition of dual, or double, criminality checks, the requirement that the underlying conduct be a criminal offense in both the issuing and executing state, but only for 32 specifically listed offense categories, and only where the conduct carries a maximum sentence of at least 3 years in the issuing state. The list runs from terrorism, human trafficking, and corruption to money laundering, cybercrime, and murder, among 26 further categories. For everything outside that list, dual criminality survives as a live requirement alongside the general 12-month threshold.

Refusal grounds are structured in three tiers. Article 3 sets mandatory grounds the executing authority must apply, including the ne bis in idem principle, the rule barring a second prosecution for the same acts. Article 4 sets optional grounds the executing authority may apply. Article 4a adds an optional ground specific to convictions rendered in absentia. The CJEU's early jurisprudence treated this list as exhaustive, a doctrinal wall the fundamental-rights case law discussed below had to work around rather than through.

The most recent fully retrieved EU-wide figures, from the European Commission's 2022 Statistics Report, recorded 13,335 warrants issued and 5,125 resulting in effective surrender, an average of roughly 57 days for a contested case. More recent aggregate figures exist in a December 2025 Council document not reviewed for this article, so the 2022 figures are used here as the most recent verified snapshot rather than an estimate.

The retained dual-criminality requirement did real, outcome-determinative work in the Carles Puigdemont litigation. Spain issued European Arrest Warrants against the former Catalan president over the 2017 independence referendum, charging rebellion and sedition, an offense outside the 32-category list. A German court held Spain's rebellion charge had no equivalent under German law and denied surrender on that specific charge; Spain withdrew the warrant rather than accept a partial surrender limited to lesser counts.

When Mutual Trust Breaks Down: Aranyosi, LM/Celmer, and the Warrant That Doctrine Hasn't Refused

Two 2013 Grand Chamber judgments set the restrictive baseline the Court would later have to work around. In Radu (C-396/11, 29 January 2013), the Court held the Framework Decision does not permit refusal on the ground that the person was not heard by the issuing authority before a warrant was issued; that right is instead safeguarded through the executing state's own procedure. In Melloni (C-399/11, 26 February 2013), the Court went further: a member state cannot use its own, potentially higher, constitutional fair-trial standard to refuse surrender, because doing so would undermine the primacy and effectiveness of EU law. The Framework Decision's refusal grounds are exhaustive. Opinion 2/13 (18 December 2014) supplied the background doctrine behind that restrictiveness: member states must presume one another's compliance with EU fundamental rights, and courts may not routinely re-verify that presumption.

Aranyosi and Căldăraru (Joined Cases C-404/15 and C-659/15 PPU, Grand Chamber, 5 April 2016) put the first crack in that presumption. Execution of a warrant must be deferred where there is a real risk of inhuman or degrading treatment, an absolute right under Charter Article 4, arising from detention conditions in the issuing state; if the risk cannot be discounted within a reasonable time, the executing authority must decide whether to end the surrender procedure. The resulting two-step test requires, first, objective evidence of a systemic or generalized deficiency, and second, substantial grounds that this particular individual faces a real risk given their specific circumstances.

Minister for Justice and Equality v LM (C-216/18 PPU, Grand Chamber, 25 July 2018), widely known by the surrendered person's name, Artur Celmer, extended the same two-step architecture from Charter Article 4 to Article 47, the right to a fair trial before an independent tribunal, in the context of documented, systemic threats to judicial independence in Poland.

Stated plainly, without softening in either direction: despite years of documented rule-of-law deterioration in Poland that the European Commission itself has formally flagged, no European Arrest Warrant issued to Poland has to date been definitively refused as a direct result of the LM test. Courts have suspended or delayed individual surrenders while assessing the test's second, individualized step, but the systemic finding alone has not yet translated into a refusal. This is a doctrine-versus-outcome gap, not a resolved question in either direction: the test is doctrinally real, and has functioned so far as a mechanism for delay rather than refusal.

Soering v. United Kingdom's extraterritorial-liability reasoning is the Strasbourg-side doctrinal ancestor of this same kind of fundamental-rights exception, though decided under the European Convention rather than under EU law. A companion article on fair trial standards under Article 6 ECHR develops that Convention-side doctrine in full; this article's Aranyosi and LM line is the parallel, but doctrinally distinct, descendant, decided by the CJEU under EU law rather than by Strasbourg under the Convention.

The European Investigation Order: Mutual Recognition Applied to Evidence

The European Investigation Order, established under Directive 2014/41/EU, does for evidence what the European Arrest Warrant does for people. It is a binding request from a judicial authority in the issuing state compelling the executing state to carry out, or transfer existing, evidence-gathering measures, replacing the older, slower mutual-legal-assistance patchwork with one comprehensive framework. Its scope covers witness hearings, telephone interception, covert investigation, banking-records disclosure, and search and seizure, channeled through a single instrument rather than instrument-by-instrument requests.

The order's clearest practical advance over the older system is procedural: the executing state must recognize an order within 30 days and carry out the requested measure within 90 days of that recognition decision, a hard, calendar-driven timeline mutual-legal-assistance requests never carried. Denmark and Ireland sit outside the system by opt-out, the same pattern recurring elsewhere in this collection: both states also sit outside the European Public Prosecutor's Office, and Denmark additionally sits outside the e-Evidence Regulation discussed below.

The distinction between the two instruments is functional, not procedural. The warrant moves a person, on a mutual-trust presumption that the preceding section shows can be rebutted only through a demanding, two-step fundamental-rights test. The investigation order moves evidence, on a fixed calendar that leaves little room for that kind of individualized, rights-based delay, with different safeguards attached to each.

Interpol: Outside the System, and Not Always Neutral

Interpol, the International Criminal Police Organization, sits entirely outside the EU legal order, a global body of more than 195 members whose central legal constraint is Article 3 of its own Constitution, which bars "any intervention or activities of a political, military, religious or racial character." Two notice channels operate with markedly different visibility: Red Notices are broadly circulated after vetting by Interpol's General Secretariat, while diffusions are circulated directly by one member state's National Central Bureau to a chosen set of others, without the same centralized publication.

A January 2026 investigation by the BBC World Service and Disclose, drawing on leaked internal files, documented extensive state abuse of both channels to target dissidents, journalists, and political exiles. Russia generated more complaints to Interpol's own complaints unit, the Commission for the Control of Files (CCF), than any other member state, and Russia, Peru, and Tajikistan held the largest active Red Notice counts as of a September 2024 snapshot: 4,817, 4,457, and 3,493 respectively. Amnesty International characterized the findings as "a grave institutional failure."

Interpol's 2026 reforms are real but recent: updated Article 3 application guidance now incorporates Article 2, respect for the Universal Declaration of Human Rights, for the first time; a mandatory online filing portal for the CCF went live in March 2026; and the CCF received increased funding for staff and IT tooling. The available baseline figures against which these reforms will eventually be measured predate them: 2024 reporting found 70 percent of the CCF's access requests and 30 percent of complaints processed late, and in 2023 the compliance-review process refused 1,603 notices and diffusions for non-compliance, 304 specifically for Article 2 or Article 3 violations. These figures describe the problem the 2026 reforms are responding to, not evidence the reforms have already reduced abuse or cleared the backlog.

The structural contrast that matters most is not any single figure, but the absence of a court. Interpol has no CJEU-equivalent body building fundamental-rights exceptions into its notice system, as with the warrant; its safeguard is an internal administrative review commission, not an independent judiciary, and its jurisdictional reach vastly exceeds the EU's own rule-of-law enforcement tools.

What Brexit Cost: UK-EU Cooperation Outside Mutual Recognition

The United Kingdom's departure from the EU legal order supplies the clearest natural experiment in this series in what falls away when a state exits the mutual-recognition system entirely. The EU-UK Trade and Cooperation Agreement, signed 30 December 2020, provisionally applied from 1 January 2021, and in force from 1 May 2021, devotes Part Three, Title VII, "Surrender," to rebuilding an extradition-equivalent mechanism from scratch, modeled closely on the pre-existing 2006 EU-Iceland/Norway surrender agreement rather than on the warrant system itself.

Two losses were structural rather than negotiable. The UK lost access to the Schengen Information System entirely, a significant operational loss since alerts in that system underlie the great majority of practical warrant-related arrests, and it lost European Arrest Warrant access outright, meaning EU states may now refuse to extradite their own nationals to the UK, a right several exercised immediately. Under TCA Article LAW.SURR.83, the default mirrors the warrant system's: a state may not refuse surrender solely because the requested person is its own national. The Article lets a state notify a formal exception, the own-nationals refusal ground, renewable every five years under Article LAW.SURR.110. Ten member states have invoked it against the UK specifically: Croatia, Finland, France, Germany, Greece, Latvia, Poland, Slovakia, Slovenia, and Sweden.

What survived comparatively well offsets those headline losses. Exchange of criminal records has shown no detrimental impact, and Prüm-framework data exchange, covering DNA, fingerprints, and vehicle registration, remains broadly effective. The UK House of Lords' own EU Committee assessed the overall picture bluntly in 2023 reporting: post-Brexit UK-EU security cooperation is "sub-optimal."

The comparative-law point here is a graduated reality, not a binary cliff-edge. Data-exchange infrastructure, Prüm and criminal records among them, proved portable through a bespoke bilateral instrument, while court-to-court instruments built specifically on EU mutual-trust doctrine, the warrant system and the Schengen Information System, did not survive exit and required a slower, more discretion-laden replacement.

What Changes on 18 August 2026, and Other Recent Developments

18 August 2026 is the date this entire architecture converges on. Regulation (EU) 2023/1543, the e-Evidence Regulation, and its companion Directive (EU) 2023/1544 together create the European Production Order and the European Preservation Order, letting a judicial authority in one member state compel a service provider located in, or offering services into, another member state to hand over subscriber, traffic, or content data directly, ordinarily within 10 days or 8 hours in emergencies. The framework enters full application across the EU, except Denmark, on that date, with non-compliance sanctions capped at 2 percent of a provider's worldwide annual turnover.

The SIRIUS project, jointly run by Eurojust and Europol since October 2017, has served as the practical knowledge base helping investigators handle cross-border e-evidence requests ahead of the framework's binding date. The date matters because it is the direct legislative answer to the most consistently repeated operational complaint across Europol and Eurojust joint reporting since at least 2018: electronic evidence sits in the cloud, in a jurisdiction unrelated to the crime, and lawfully obtaining it has historically taken too long.

Regulation (EU) 2024/3011, in force since 7 January 2025 but not applicable until 1 February 2027, sets common EU-wide rules for moving an entire criminal proceeding, not just a person or an item of evidence, to whichever member state is best placed to investigate or prosecute, explicitly to prevent parallel prosecutions of the same suspect and facts across multiple states. Its detention-credit provision requires the receiving state to deduct, from any eventual custodial sentence, all detention time already served in the requesting state in connection with the transferred proceeding.

This same 2024 Regulation connects directly to the European Supervision Order, established under Framework Decision 2009/829/JHA of 23 October 2009, which lets a non-resident EU citizen charged in one member state be supervised in their home member state using non-custodial measures rather than remaining in pretrial detention abroad. This article's concern is the order's mutual-recognition mechanics; a companion article on pretrial detention limits in democratic systems documents its real-world record, a near-dead letter in Italian practice specifically, in more depth than is repeated here.

Frequently Asked Questions About Cross-Border Criminal Justice Cooperation

Does Europol function as a European equivalent of the FBI, making its own arrests?

No. Europol carries no arrest power and no independent investigative authority; its role is intelligence-sharing and coordination, and national police, acting under national law, make every arrest in an operation Europol supports.

Did the European Arrest Warrant do away with the requirement that the underlying conduct be criminal in both the issuing and executing state?

Only for a defined set of offenses. Dual criminality was abolished for 32 listed categories carrying at least a 3-year sentence threshold; outside that list, the requirement remains, which is precisely what stopped Spain's surrender request against Carles Puigdemont on the rebellion charge.

Now that the CJEU has built a fundamental-rights exception into the warrant system, does a member state with documented rule-of-law problems simply have its warrants refused?

Not so far, as detailed above: as of 2026 the Aranyosi and Căldăraru/LM test has yet to produce a single definitive refusal of a Polish warrant, despite years of documented concerns over judicial independence there. In practice it has supported delay rather than refusal.

Does Interpol answer to the same fundamental-rights oversight that governs the European Arrest Warrant?

No. Interpol is a separate global organization of more than 195 members, entirely outside the EU legal order, and its sole internal constraint is a constitutional non-intervention clause enforced by an administrative review commission rather than a court. A January 2026 investigation documented member states using its notice system against dissidents and political exiles.

Related Articles

This article sits alongside companion pieces examining adjoining parts of the same architecture from different angles. A companion article on fair trial standards under Article 6 ECHR develops the Strasbourg-side doctrinal ancestor of the CJEU's fundamental-rights exception discussed above, tracing Soering's extraterritorial-liability reasoning through the European Convention system rather than EU law. A companion article on pretrial detention limits in democratic systems covers the individual-liberty and alternatives-to-detention framing of the European Supervision Order this article treats mechanically above. A forthcoming article on police accountability and civil rights protections will take up the institutional oversight questions raised in passing here by the Bundeskriminalamt and GTAZ. Readers seeking Europol and Eurojust's cybercrime-specific operational statistics should consult this workspace's article on cybercrime legislation across EU jurisdictions.