#4: Checking the checkers: how much oversight does the defence really have over privilege filtering?

In this fourth blog, we consider what oversight the defence has under the new regime and why that oversight falls structurally short.

From judicial safeguard to prosecutorial trust

The Dutch Supreme Court could not have been clearer: legal privilege is undermined the moment the prosecutor leading the investigation gains access to material that is, or may be, privileged. That is why the Court held that any filtering exercise requiring access to potentially privileged material must be carried out by, or under the responsibility of, the investigating judge.

The legislature disagrees. Under the bill, it is not the investigating judge but the public prosecutor who takes charge of the filtering process. Handing that much trust to the prosecution ought to come with a corresponding increase in oversight. That is precisely where the new regime falls down.

Why 'reasonable suspicion' matters so much

Before turning to the oversight mechanisms, it is worth pausing on the 'reasonable suspicion' test, which now sits at the heart of the new regime. As we explained in our first post, this test marks a real departure from existing case law. It no longer asks whether privileged material is likely to be mixed in with the bulk data, but it asks whether a specific document within that bulk is privileged. That holds true, in the legislature's own words, even where "there is a substantial likelihood that privileged material remains among the data that has not been filtered out."

What oversight does the defence actually have?

Under the current system, the investigating judge can let the defence check that no privileged material has slipped through the filtering process (the courts have published a formal Working Method for filtering digitally privileged material for exactly this purpose). The defence can also lodge a formal complaint about how the filtering and selection of privileged material was carried out.

So, what does the bill offer instead?

The defence can lodge a complaint against access to privileged material (Article 6.4.11a et seq.), against seizure (Article 6.4.1), and against the taking or ordering of data (Article 6.4.5). Where the investigating judge rules that privileged material may be accessed, the privilege holder can also appeal that decision to the district court and, ultimately, to the Supreme Court (Article 2.7.71).

At first glance, that looks like adequate oversight. It is not. The new 'reasonable suspicion' test, the same one we flagged in our first post, narrows the scope of that oversight considerably, and how far the defence can push it depends heavily on the effort (read: money) it can throw at the problem.

Complaints about access: too little, too unappealing

Articles 6.4.1 and 6.4.5 of the first implementation act already allow complaints against seizure and against the taking or ordering of data. The legislature identifies, however, that these provisions do not cover complaints about the use or accessing of that material. To plug that gap, it has introduced a separate complaints route specifically for access to privileged material (Article 6.4.11a). The starting assumption is that material which may not be accessed is retained rather than destroyed (see our third post for why that matters).

It is a solution to a problem nobody has. As we argued in our third post, it is hard to imagine a suspect who would rather see privileged material retained than returned or destroyed outright. Worse, this complaints route is made deliberately unattractive to use. Because its sole purpose is to stop or prevent further access, its scope is narrow: a complaint can only be brought over access to privileged material that is "imminent or has already taken place", in the legislature's own phrasing. Once that access has ended, or once the prosecutor simply states that no access will be taken, the legislature considers there to be no remaining grounds for complaint, and the complainant will be found inadmissible. Whether this complaints procedure will see much use in practice remains to be seen.

Complaints about seizure: what you see is what you get

Complaints against seizure and the taking or ordering of data do not face these particular obstacles. But under the new definition of 'reasonable suspicion', the defence still must point to specific documents it believes are privileged. If it cannot do that yet, it is left relying entirely on the judgement and diligence of the public prosecution service, the very body running the filtering process.

The filtering process itself: a black box

That filtering process is a black box for the defence. Only after the following three conditions are satisfied will the investigating judge have to give the privilege holder concerned the opportunity to make its position known, namely:

1. the investigative authorities encounter documents in respect of which a reasonable suspicion exists that legal privilege extends to them;
2. they nonetheless wish to use those documents; and
3. the documents are for that reason submitted to the investigating judge.

In short, this procedure only addresses the output of the filtering and selection that preceded it.

Nothing in the bill lets the defence test whether that preceding filtering and selection was done correctly and completely (see our second post on this point). The reporting duty the legislature points to is nowhere near enough. Yes, it requires the selection process to be logged in an official record, but only the fact that a selection took place, the method used ("automated or otherwise"), and whether any material was flagged as potentially privileged. What is missing is any record of why material was not flagged as potentially privileged. That is precisely the point the defence has every reason to want to check. And that is before we even ask whether investigators have the expertise to make that call correctly in the first place, a question we would answer with an unambiguous no.

Deep pockets: cost and delay as a barrier for the defence

Where large volumes of digital data are involved, the defence can try to get a genuine grip on the filtering process by running its own filtering exercise on a copy of the seized data. That would let it point to specific privileged documents far more precisely when lodging a complaint.

The trouble is that this kind of filtering requires specialist software, which becomes expensive fast once datasets grow. It also demands significant staff time, drafting search terms, reviewing results, which adds further cost on top. The risk is obvious: a suspect without deep pockets ends up in a materially worse position than one who can afford to run a parallel review.

Thorough filtering of large datasets also simply takes time. While investigators are already deep into their inquiry, and gaining access to privileged material day by day, the defence is often still setting up its own filtering exercise. Every time it comes across privileged material, it must flag this to the investigation team, which itself sits uneasily with the protection privilege is meant to provide. If the response is unsatisfactory, a complaint is the obvious next step. It is not hard to imagine this producing a steady stream of complaints, each requiring the courts to rule, document by document, on whether access should be allowed.

Where this leaves us

The legislature is handing the filtering process to the public prosecutor, but the oversight built in to compensate for that shift does not go far enough. The complaints route for access to privileged material is inadequate and unattractive to use. The complaints route for seizure assumes the defence can already point to specific documents, which in practice it usually cannot. Appeals against the investigating judge's decisions are limited to documents that the investigators themselves flagged as potentially privileged. Anything outside that scope escapes scrutiny altogether. A suspect who genuinely wants to hold the system to account has little choice but to run a parallel filtering exercise of their own, with all the cost and delay that entails. In practice, legal privilege ends up depending on the suspect's financial means.

In summary

With this fourth blog, we conclude our series on the second supplementary act to the new Dutch Code of Criminal Procedure. Throughout the series, we have argued why the bill fundamentally falls short, on several interrelated points, in protecting legal privilege. Our main objections and recommendations can be summarised as follows.

First, because the bill allows access to bulk data before any protection kicks in, a breach of privilege cannot be undone once it happens. What has been seen cannot be unseen.

Second, filtering privileged material should never sit with the public prosecutor, who is by definition the prosecuting party and therefore not the right body to be screening potentially privileged material. A better solution lies in structural investment in the judiciary specialist software and a dedicated bench of investigating judges for privilege disputes.

Third, destruction should remain the default for material found to be privileged. If the legislature is worried that this could make material the defence needs otherwise inaccessible, the obvious answer is to place responsibility for retention with the defence itself, or with the investigating judge not with the prosecution.

Fourth, the bill does not provide any procedural safeguard that lets the defence test the filtering process itself. If this framework goes ahead, it needs an effective way to check that process one that does not depend on how much a suspect can afford to spend.

The problem the legislature is trying to solve is real and urgent. The proposed solution, however, falls well short on rule-of-law grounds. Legal privilege is not a case-management inconvenience to be traded off for efficiency. It is a fundamental right, and it does not bend to the demands of an investigation.