“…the court now has a very salutary power, both to be exercised in a claimant's favour or, where appropriate, in a defendant's favour. It enables the court to dispose summarily of both claims or defences which have no real prospect of being successful. The words "no real prospect of being successful or succeeding" do not need any amplification, they speak for themselves. The word "real" distinguishes fanciful prospects of success or, as Mr Bidder submits, they direct the court to the need to see whether there is a "realistic" as opposed to a "fanciful" prospect of success.”

Rule 8(3)(c)

The Tribunal may strike out the whole or a part of the proceedings if—

the Tribunal considers there is no reasonable prospect of the appellant's case, or part of it, succeeding.

determining whether the information is held;

locating the information, or a document containing it;

retrieving the information, or a document containing it; and

extracting the information from a document containing it.

It is well established that section 12 FOIA does not oblige public authorities to keep their records in such a way that they can be quickly and easily located. The costs limit under section 12 is considered on the basis of the public authority's actual record keeping practices, not on the basis of how the requested information should have been kept (Commissioner of Police for the Metropolis v Information Commissioner and Mackenzie [2014] UKUT 479 (AAC)).

“NHSE has repeatedly and in detail set out the facts and assumptions upon which its estimate is based. The ability to neatly use technology to respond to the request, which Mr Ryan assumes exists, in fact does not. It is no part of this Tribunal’s jurisdiction to make any assessment about NHSE’s practices, procedures and record management techniques more generally. It is clear that the scope of Mr Ryan’s request is such, that to attempt to comply with it would result in the appropriate limit being vastly exceeded.”

Firstly, we must assess the weight that we give to the ChatGPT evidence. We place little weight upon that evidence because there is no evidence before us as to the sources the AI tool considers when finalising its response nor is the methodology used by the AI tool explained. If comparisons are drawn to expert evidence, an expert would be required to explain their expertise, the sources that they rely upon and the methodology that they applied before weight was given to such expert evidence. In the circumstances we give little weight to the ChatGPT evidence that searches should have been conducted in the form set out within that evidence.”

As is the practice in a case in which the public authority has informed the complainant that it holds the information, the Commissioner asked the Cabinet Office to provide a detailed estimate of the time/cost taken to provide the information falling within the scope of this request.

In its internal review response and its submission to the Commissioner the Cabinet Office stated that it had interpreted the term ‘document’ to include emails, word documents, spreadsheets and presentations. It also stated that it had also interpreted ‘involving’ to mean documents on the subject of wildfire preparedness which the Cabinet Office created, has had some input into or on which the Cabinet Office has been consulted.

The Cabinet Office informed the Commissioner that an initial search was carried out by an official within the policy team believed to hold information in scope of this request for the period 1 January 2019 to the date of the request. It was established that the term ‘wildfire’ generated 3,568 emails. These emails, however, covered a wide range of subjects such as invitations to meetings, risk assessment and briefing notes. Several of these emails were also shared for information purposes, and therefore would not fall within the relevant subject matter for this request.

The Cabinet Office has estimated that it would take an average of 30 seconds for an official to determine whether an email would be within the scope of this request, to isolate this email and save it to a separate area if appropriate. It estimated that this stage would take approximately 29 hours (3568 emails x 30 seconds).

In relation to documents, the Cabinet Office stated that a search was also carried out within the main file repository for emails held by the policy team for the search terms ‘wildfire’ and ‘preparedness’. It was, however, not possible to limit the results of these searches for the period 1 January 2019 to date as the dates of these files were overwritten during the transition of the system to using Google products. These searches returned 1,687 returns, also covering a wide range of records. The Cabinet Office estimated this to also take an average of 30 seconds for an official to access and isolate each record. It estimated that this stage would take approximately 14 hours (1687 x 30 seconds).

The Cabinet Office therefore concluded that conducting searches through for the terms that concern the scope of the request over 3,568 emails and 1,687 documents, as well as determining which records are in scope, would require approximately 43 hours to accomplish.

The Commissioner considers that the Cabinet Office estimated reasonably that it would take more than the 24 hours to respond to the request. The Cabinet Office was therefore correct to apply section 12(1) of FOIA to the complainant’s request.”

The Appellant has provided screen shots of AI suggested search paths that the public authority might consider in interrogating it systems. My reading of those suggestions also provides the Appellant with ways he might refine his information request.

It seems to me that the Tribunal might be assisted by the public authority submissions on its reliance on section 12 FOIA. It would be a matter for the public authority as to whether the AI suggested search pathways should be considered or any submissions on that point.

At this stage I do not consider the appeal has no reasonable prospect of success pursuant to Rule 8(3)(c).

The Applicant’s views on methods of search based on artificial intelligence have no reasonable prospect of showing error of law in the s 12 decision, not least because no error of law ground is advanced, but additionally because of the principle summarised by the FTT in Arthur (see above) and in Ryan.

In line with Oakley, little weight should be attached to the Applicant’s assertions in reliance on responses to “quick requests of AI chatbots”: Grounds, p 3.

For completeness and to assist the Tribunal with the evidential issue posed by the FTT in its 24 April 2024 decision (see above) the Cabinet Office has filed and served with its application the witness statement of David Canning, dated 1 May 2024, confirming that it does not have AI tools that enable it to search its records in the way that the Appellant envisages.

DISCUSSION