Section 31: Show Us the Missing Link

Section 31 of the Freedom of Information Act is increasingly becoming the central issue in our national facial-recognition investigation.

Across the requests we are tracking, police forces have repeatedly cited Section 31 — Law Enforcement when considering whether to disclose information about facial-recognition systems.

In a growing number of cases, forces have subsequently provided the same more detailed formulation:

Section 31(1)(a) and (b)
Questions 5(a), 5(b) and 6
“Prejudice to tactical capability”

The explanation which follows is also remarkably consistent: disclosure would make the facial-recognition tool less effective in its primary function of preventing and detecting crime and apprehending offenders.

At the time of writing, 14 police forces have supplied that same full limb, scope and prejudice formulation.

But Section 31 is not engaged simply because information relates to policing, facial recognition or something which might broadly be described as a “tactical capability”.

It is a prejudice-based exemption.

That distinction matters.

For Section 31 to apply, there must be a meaningful connection between disclosure of the particular information requested and the harm claimed.

The Information Commissioner’s guidance asks authorities to identify the protected law-enforcement interest, establish that the harm is real, actual or of substance, demonstrate a causal link between disclosure and that harm, and address the likelihood that the harm would actually occur.

Only after the exemption is properly engaged does the separate public-interest balancing exercise arise.

Read the ICO Section 31 prejudice-test guidance.

What is the missing link?

Our request was deliberately written to avoid operationally sensitive material.

We did not ask for watchlists, details of individual investigations, covert tactics, security credentials, exploitable technical vulnerabilities or network architecture.

Yet the repeated response we are receiving is:

Disclosure → prejudice to tactical capability → facial recognition becomes less effective.

What is frequently missing is the step in between.

How?

How would disclosure of the particular records we requested cause that harm?

It is not enough simply to attach the words “tactical capability” to information and move directly to Section 31.

And there is a particularly relevant ICO decision which demonstrates why.

IC-333147-L9X3 — North Wales Police

On 4 March 2025, the Information Commissioner issued decision notice IC-333147-L9X3, concerning a Freedom of Information request to the Chief Constable of North Wales Police.

The request concerned the deployment of Live Facial Recognition at Holyhead.

North Wales Police relied on Sections 31(1)(a) and (b) — prevention or detection of crime, and apprehension or prosecution of offenders — to withhold information contained in a written authority for an LFR deployment.

The Commissioner did not decide that Section 31 was irrelevant to facial recognition.

Quite the opposite.

He accepted that a limited amount of genuinely operational information could properly be withheld. The decision accepted protection for material capable of revealing operational deployment detail and information of value to people seeking to evade policing activity.

But that was only a small part of the material.

For the majority, the Commissioner reached a very different conclusion.

North Wales Police had not demonstrated the necessary causal link between disclosure and the claimed law-enforcement harm. The Commissioner found that Sections 31(1)(a) and (b) had effectively been applied in a blanket fashion to the document, and required disclosure of the information for which the exemptions were not engaged.

Read ICO decision IC-333147-L9X3.

That distinction is critical.

Specific operational information may be capable of satisfying the prejudice test.

The fact that a document concerns facial recognition does not mean that every part of that document does.

The exemption has to fit the information.

And the causal link has to be demonstrated.

Then something changed

Until today, the repeated wording appearing across police forces presented us with an interesting evidential problem.

Fourteen forces have now supplied the same substantive explanation:

s31(1)(a) and (b)
Q5(a), Q5(b) and Q6
prejudice to tactical capability
the tool would become less effective at preventing/detecting crime and apprehending offenders

That was strong evidence of common wording, precedent material or a shared template.

But similarity alone could not tell us where the wording came from.

There are perfectly ordinary explanations for some shared language. Police forces collaborate. Some share information-management teams. Some use shared legal services.

Without evidence, we were not prepared to claim more than the documents established.

Today we received new information.

Merseyside Police

After Merseyside Police missed its original response deadline, we asked what had happened.

Its response included this:

“Your request is currently sitting with the national Central Referral Unit for further advice & guidance to be provided to all police forces.”

That is considerably more informative than a generic reference to stakeholder consultation.

It establishes that Merseyside’s handling of this request has been referred to a national Central Referral Unit, and that advice and guidance arising from that process is intended to be provided to police forces nationally.

It does not establish what that guidance says.

It does not establish that the Central Referral Unit wrote the repeated Section 31 formulation.

And it does not prove that forces have been instructed to respond in a particular way.

But we now know that national-level advice is involved.

Then Derbyshire replied.

Derbyshire Constabulary

Derbyshire had also missed its original deadline.

In response to our follow-up, Inspector Aaron Shaw explained:

“due to the sensitive nature and the depth of the request it is being reviewed by the National Central referral Unit who provide a national steer on FOIs.”

Derbyshire therefore independently confirms the involvement of the same national unit.

More importantly, it describes its role:

providing a national steer on FOIs.

We now have two separate police forces documenting national involvement in the handling of this request.

Merseyside says the unit is providing “advice & guidance to be provided to all police forces.”

Derbyshire says it provides a “national steer on FOIs.”

That is new evidence.

What we know — and what we don’t

The evidence now supports several statements.

A substantial number of police forces are considering Section 31 in response to the same facial-recognition FOI request.

Fourteen forces have supplied the same detailed limb, question scope and prejudice formulation.

Two forces have now independently confirmed that the request is being considered by a national Central Referral Unit which provides national FOI guidance or steer.

What we cannot yet say is that the Central Referral Unit created the repeated wording, instructed forces to use Section 31, or predetermined the outcome of their individual public-interest tests.

Those remain questions.

And they are now questions worth asking.

Because ultimately the legal test does not change according to how many forces receive the same guidance.

For every piece of information withheld under Section 31, the question remains:

What law-enforcement harm would disclosure cause?

What is the causal link?

The ICO has already considered that question in the specific context of Live Facial Recognition.

It accepted protection for genuinely operational material.

It rejected the blanket application of Section 31 to wider information where the necessary connection to prejudice had not been demonstrated.

That is not an argument against Section 31.

It is the test.

Show us the missing link.

Trust requires evidence.