Europe’s criminal justice runs at two speeds — and the slower one carries the rule of law
By Alexis Anagnostakis
A client is arrested at a border crossing on a European Arrest Warrant. Months later, a court in the executing State refuses surrender: the facts are stale, the prosecution abroad is at investigative stage only, detention conditions in the issuing State are documented as inhuman. A reasoned judicial decision, delivered by a court of an EU Member State, applying EU law and the Charter.
He is released. He travels. He is arrested again three weeks later, in a different Member State, on the same warrant.
Nothing has gone wrong procedurally. That is the point. The system is functioning exactly as designed — and the design is the problem.
The pattern that nobody names out loud
Anyone who has followed the recent institutional debates on the future of EU criminal justice will have noticed something that never appears as an agenda item, because it is not a topic. It is a habit.
When the beneficiary of a proposed rule is an enforcement authority, the EU legislates in the imperative. Regulations. Directives. Directly applicable instruments with deadlines, transposition duties, infringement proceedings, and the full apparatus of judicial enforcement behind them. Access to data, retention, interception, cross-border evidence gathering, expanded agency mandates, a prosecutorial office with direct powers across borders — all of it hard law, and much of it moving quickly.
When the beneficiary is the individual facing the State, the register changes. Recommendations. Guidelines. Soft-law measures. Further study. Exchange of best practice. A commitment to keep the matter under review.
Pre-trial detention is the cleanest example. There has been a non-binding Recommendation in place since 2022. Prison overcrowding across the Union has not improved; monitoring bodies continue to report that no Member State fully meets European detention standards in practice; the deprivation of liberty before conviction remains, in too many jurisdictions, the working default rather than the exceptional measure the law says it is. The response, after four years of demonstrated ineffectiveness, has been a proposal for more soft law.
There is a legal basis for binding minimum rules. Article 82(2)(b) TFEU is not ambiguous. What is absent is not competence. It is appetite.
A Union that legislates bindingly for its own powers and advisorily for its citizens’ rights is not balancing interests. It is expressing a hierarchy.
Why this is a rule-of-law question, not a defence lobby question
The instinctive reply is that defence lawyers would say this. We represent one side; of course we want more rules on our side of the table.
That reply misunderstands the structure of the argument.
Mutual recognition rests on mutual trust. But mutual trust, in EU law, is not a sentiment. It is a legal presumption — the presumption that a judicial decision from any Member State was produced by a system that respects fundamental rights, and therefore deserves to be executed almost automatically, with minimal review by the receiving court.
A presumption of that weight is only defensible if the thing presumed is actually secured. Not aspired to. Secured — by binding standards, enforceable before courts, uniform enough that a judge in Athens can rely on what a judge in Bucharest or Dublin was obliged to do.
Where the underlying protection is soft law, the presumption has nothing underneath it. It stops being trust and becomes assumption. And an assumption operating with the force of law is precisely what the rule of law exists to prevent.
This is why defence rights are not a counterweight to mutual recognition. They are its load-bearing structure. Every unfilled gap in procedural safeguards is a crack in the foundation of the enforcement architecture that Member States themselves want to expand. The people who should be most alarmed by soft-law procedural rights are, on a rational analysis, prosecutors.
“The case law will handle it” — a promise the case law cannot keep
The prevailing position among Member States is that no legislative revision of the older mutual recognition instruments is necessary, because the Court of Justice has developed a fundamental rights exception through its case law, and this has proved a sufficient corrective.
From the courtroom floor, this is not how it looks.
Case law is retrospective. It corrects a violation that has already occurred, to a person who has already been detained, in proceedings that have already cost them their liberty, their employment, and frequently their family life. The remedy arrives years after the harm. Legislation prevents; litigation compensates. These are not interchangeable.
Case law is unevenly available. Taking a point to Luxembourg or Strasbourg requires a lawyer who recognises it, a client who can fund it or a legal aid system that will, a national court willing to refer, and a tolerance for years of delay. The clients who most need the protection are structurally the least able to generate the case law that would provide it. Rights that exist only for those who can litigate them are not rights of general application.
Case law does not bind the practice of authorities in the way a statute does. A defence lawyer cannot walk into a first-instance hearing in a busy provincial court and plead a jurisprudential trend with the same effect as a transposed directive. Judges apply what is in front of them in the code.
And case law is not a plan. It cannot recast an instrument, harmonise a definition, create a consultation procedure between authorities, or build an alternative to detention. Courts resolve disputes. They do not design systems. Asking the judiciary to do the legislature’s work is not restraint — it is the transfer of a political decision to an institution that never sought it and cannot properly discharge it.
The European Parliament called for legislative action on the arrest warrant framework more than a decade ago. That call is still unanswered. “The case law will handle it” has now had ten years to prove itself.
Three tests for the coming cycle
Advocacy is easy to produce and hard to measure. So here are three benchmarks that are concrete enough to be failed.
Test one: does a refusal mean anything? When a court refuses surrender on grounds that are EU-wide in nature — ne bis in idem, a real risk of treatment contrary to Article 19 of the Charter, a flagrant denial of justice, disproportionality — that decision should bind every Member State, not just the one that issued it. Anything less turns free movement into a trap and invites the applicant State to shop for a more compliant forum. It should follow that once such a refusal exists, the corresponding alert cannot continue to circulate in the Schengen Information System, or migrate into an INTERPOL notice or diffusion, with an effective remedy where it does. Recognition of refusals is the single cheapest structural repair available. Its absence is not an oversight; it is a choice.
Test two: is proportionality assessed where the damage is done? Proportionality review at the executing stage is damage limitation. The person has already been arrested. The meaningful test is at issuance, based on the sentence realistically in prospect rather than the statutory maximum, with a duty to consider whether an investigation order or a supervision order would achieve the same objective. An arrest warrant used to secure an interview is a coercive instrument deployed for a communicative purpose.
Test three: is technology used to reduce coercion, or only to increase reach? Video-conferencing has been discussed almost entirely as an efficiency question. Reframe it as a liberty question and it looks different: a mandatory video-link hearing shortly after arrest, allowing the issuing authority to see the person, hear them, and decide whether the warrant is still needed or an alternative will do. Rules permitting a suspect resident in another Member State to be heard remotely, so that arrest is not the only mechanism for securing attendance.
The safeguards must travel with it — informed and express consent, an unqualified right to be physically present at trial, interpretation, verified identification, genuinely confidential lawyer–client channels, and equivalent technical quality for both sides of the case. Virtual presence is not presence. But used deliberately, the video link is one of the few instruments capable of reducing detention without asking any Member State to surrender an inch of jurisdiction. That makes it the most politically achievable liberty reform on the table, which is a reason to press it rather than to treat it as a minor logistical matter.
In the meantime
None of this excuses waiting. Until the framework is repaired, the burden sits where it always has: on the practitioner who raises the point that nobody expects, in a hearing where everybody is in a hurry.
Ask for the case file in the issuing State, in writing, and document the refusal. Raise proportionality at issuance even where the local court insists it has no competence to consider it. Propose the supervision alternative on the record, so that its rejection becomes a reviewable fact. Challenge the alert as well as the warrant. Preserve the point for appeal even when the appeal seems hopeless — the case law that Member States now cite as a sufficient corrective exists only because lawyers made arguments that were, at the time they were made, considered unarguable.
The gap between what the law promises and what the client experiences is not filled by institutions. It is filled, case by case, by people willing to stand up and say that this is not what the Charter says.
The standard by which this will be judged
Efficiency is a legitimate objective of a justice system. It is not the objective. A system can be fast, interoperable, digitally integrated, and thoroughly unjust — indeed, speed and integration make injustice more efficient too.
The measure of the next institutional cycle will not be how many instruments are adopted or how quickly evidence crosses a border. It will be whether the register changed: whether, for once, the rights of the individual were written in the same imperative mood as the powers of the State.
Until then, we should be precise about what we have. Not a shared area of freedom, security and justice — a shared area of enforcement, with justice recommended.
The views expressed here are personal and do not represent the position of any organisation with which the author is associated.
