+44 (0)20 7353 6000
25 May 2023

Could a change of disclosure regime help the SFO’s disclosure headache

On 10 March, the Serious Fraud Office (SFO) dropped its prosecution of two G4S managers whom it had accused of defrauding the Ministry of Justice in relation to a major electronic tagging of offenders contract. Whilst it was coy about why it had to collapse its own case it is clear that the principal cause of this fiasco is that, despite years of effort, ultimately the SFO was unable to fulfil its statutory Criminal Procedure and Investigations Act 1996 (CPIA) duty to disclose relevant unused material to the defence. This inability sank its case. This disclosure regime proved impossible to satisfy even for the law enforcement agency which customarily accumulates vast amounts of electronic material in the course of its investigations.

This fiasco is not the outcome of exceptional circumstances. It is the latest episode of a systemic or institutional problem which has been afflicting and undermining SFO cases for many years. We know about its prosecutions that fail because of unmet demands imposed by this regime. The doomed Unaoil and Serco cases are other recent examples. Those two defeats led to the Altman and Calvert-Smith reviews respectively, and both of them laid bare the disclosure-related blunders committed in those cases.

The problem is greater than these public defeats. We know little about the SFO investigations that are aborted principally because of anticipated disclosure problems before they even progress to the prosecution stage; such investigations may well be terminated even though the evidence of criminal conduct is sufficient. Overall, the CPIA regime, when it comes to the prosecution of the type of cases that the SFO exists to prosecute, is a millstone around its neck.

The reaction of those responsible for superintending the SFO, for equipping it with what is needed to enable it to prosecute cases to conviction, has always been to deny that there is a fundamental problem inherent to the CPIA regime. To sustain this myth in the aftermath of another case collapsing, they implicitly blame individual prosecutors, contending that the root of the recent debacle is not the regime itself but how it was applied. In other words, there were inexplicable errors committed by the case team. Alternatively, when the single point of failure explanation is untenable because of a concatenation of errors, to assure that the problem will be fixed following a warts-and-all review or that it has already been fixed as a result of recent robust guidance handed down by the Court of Appeal as to what the regime requires (i.e. does not require) of a prosecutor.

This litany of hand wringing has persisted too long whilst disclosure has been allowed to become the SFO’s Achilles Heel in many of its cases. The loss of the trio of above cases in as many years points to a major problem. The question is: when will the superintendents realise, or be willing to admit, that they have been flogging a dead horse? There has been no lack of endeavour to tinker here and there to attempt to make the CPIA regime fit for purpose. It was implemented in 1997 and problems did not take long to emerge. Beginning with the Auld Review that reported in 2000 and even then recommended major reform, there has been if not an abundance than a plethora of subsequent reviews, protocols, codes of practice and Court of Appeal Criminal Division (CACD) judgments that have added layer upon layer of mandate or guidance all purporting to help ensure the regime operates effectively and fairly. Law enforcement has been inundated with such material. Taking the AG’s guidelines on disclosure as an example of this febrile cottage industry, their first edition was promulgated in 2000. By now if the annexes and supplements to them are included, they have reached their tenth edition. The regime is one that relies on so much scaffolding to keep it standing.

The status quo has many staunch defenders. They, of course, have invested heavily in trying to salvage the CPIA regime. They are united in trenchant opposition to an abandonment of this regime and its replacement by the one evolved by the common law that immediately preceded it. They contend that a return to the common law would be retrograde, a reinventing of the wheel and a distraction. In relation to the types of case prosecuted by the SFO, were the common law regime to be restored it would entail major upheaval. That regime entails that the SFO would make available to the defence the vast majority of the material that had been gathered during its investigation. Probably tetrabytes of data would be disclosed. The defence would then become the party responsible for surveying this amorphous mass on behalf of the accused. The restoration opponents disparagingly call this regime the giving to the defence “the keys to the warehouse”. This is a vivid but fair metaphor for what would happen if the common law regime was restored.

The retired Court of Appeal judge Sir Peter Gross was an outspoken opponent of restoration and his opinions still rightly command attention. They are cited with approval today when restoration is mooted. It is appropriate therefore to consider his reasoning. Sir Peter reviewed the CPIA regime in 2011 and again in 2012 at the request of the then government and has given speeches about this subject since. When, during 2018, the Justice Select Committee considered whether the regime should be reformed it is apparent from its resulting report that it was swayed by his opinion as to why the keys to the warehouse regime would be inferior to the CPIA one. Consequently, it too came out against fundamental change. It concurred with the conventional wisdom that what was needed to avoid further disclosure causes célèbres was additional “clear guidance” and “greater clarity” from the AG in the forthcoming edition of his guidelines. These were promulgated during 2020 but, plainly, they did not avert the trio of SFO defeats mentioned above.

Why did or does Sir Peter oppose a return to the keys to the warehouse regime? This is what he wrote about it in his 2011 review:

“Keys to the warehouse: Mention has been made from time to time of the ‘keys to the warehouse’ approach… With respect to the observations made to us in favour of the keys to the warehouse and which we have carefully considered, we are not attracted to it. Our reasons are these:

  1. As already underlined, it is necessary to guard against any proposal which simply transfers cost from one public purse to another. To an extent, the keys to the warehouse approach does just that. Adopting this approach would or should lead to a reduction in prosecution costs; but, conversely, it is also likely to result in an increase in legal aid costs. Unlike the position in the US – where, as we understand it, much white collar defence work is privately funded – the expectation here must be that defence costs will be funded by legal aid. …
  2. Matters do not rest there. In principle, we find it difficult to see how a diligent prosecutor could rest content with the keys to the warehouse approach, without wishing to familiarise himself/herself with the same material – thus posing the risk of duplication. Quite apart from the need to comply with the prosecution’s CPIA regime disclosure obligations, a diligent prosecutor would wish to ascertain how such material affected the strengths or weaknesses of the prosecution case. An element of abdication might otherwise be involved…

In a lecture to the CBA delivered during June 2018 he said: –

Similarly, I am strongly opposed to the ‘keys to the warehouse’ approach. It would increase the pressures on limited resources and result in the duplication of effort. Any diligent prosecutor would want to look at the material before handing it over to assess its impact on their case. The ‘keys to the warehouse’ in an overstretched system is simply not viable. At most, it transfers the problem without solving it.”

Are his objections to the warehouse model persuasive? The kernel of them is that transferring the responsibility for examination of the unused material from the prosecution to the defence is either cost neutral in that it “transfers cost from one public purse to another” or would cost more because it would “result in the duplication of effort”.

Insofar as typical SFO cases are concerned (Sir Peter was not specifically concerned with them but with the normal diet of criminal law litigation) his views can be criticized. First, whilst this is not universal, the majority of accused are not legally aided but enjoy the advantage either of private means or third-party funding such as from their actual or erstwhile employer or D&O insurance. So whilst Sir Peter is correct to mention the transfer of cost, he is mistaken in believing that it must all ultimately come from public funds. In fact, shifting the responsibility for the inspection of the unused material to the defence will probably result in a substantial reduction in the SFO’s annual expenditure and an overall saving to the public purse. It would be worthwhile trying to establish how much the SFO spends annually on its CPIA related disclosure duties.

Second, concerning duplication of effort, it is unclear what is the vice that Sir Peter envisages. It is, after all, a wise premise of adversarial litigation that “he/she who knows only his side of the case knows little of that” and to an extent both sides will want to examine the unused material albeit at different moments. Generally by an SFO investigator or lawyer when seeking to find material that supports or undermines a putative prosecution. Later post charge by a defence lawyer when trying to find substance for their client’s instructions. This involves duplication but it is not a vice. Indeed, freed from the onerous burdens of having to review and schedule the mass of material and identify what may pass the disclosure test pre and post service of a defence statement, the SFO team will inevitably spend far less time examining the unused material than it does now.  Potentially a huge saving in SFO resources could be achieved. Thus, restoration is very unlikely to vindicate Sir Peter’s claim that it would “increase the pressures on limited resources”.

Third, whilst this is not conveyed in the above extracts of what Sir Peter has argued, a constant theme of his advocacy was that extra resources are needed to make the CPIA regime work coupled with optimism that they will be forthcoming. In other words, because the cost of failure is so high the government will provide extra money for the SFO and other prosecutors. Unfortunately, this pious hope is naïve. The prospect of significant additional money being pumped into the SFO to facilitate fulfilment of its CPIA related duties is a mirage. Such financial reinforcement will not ride to the rescue.

Fourth, another theme mentioned by Sir Peter that has been taken up enthusiastically, especially in recent years as AI has advanced, is the optimistic expectation that technology will soon be able largely to perform the CPIA disclosure duties. This apparent panacea for the endemic problems is just around the corner. Whilst it must be true that AI will be able to assist and perhaps hugely assist in alleviating the drudgery and burdens that reviewing unused material in an SFO case inevitably attracts, it is unlikely to supplant the human mind wholesale when it comes to applying the disclosure test. Furthermore, we must not forget that technology was at the heart of two recent and huge miscarriages of justice: the Post Office scandal and the Robodebt one in Australia. Both show that where technology is over relied upon and goes wrong, the scale of the injustice can be unprecedented.

Whilst Sir Peter did not ventilate them, from other quarters there are two additional objections to the keys to the warehouse model which ought to be considered.

First, data protection, in particular sensitive personal details of an individual being disclosed to the defence if they have free access to the proverbial warehouse. This is a danger difficult to guard against and impossible to eliminate. However, in the context of an SFO case where the unused material will predominantly or entirely be business records, the risks are far less than for a case of domestic abuse or sexual violence. Moreover, it is unlikely that the risk of violation of someone’s privacy would deter a complaint of serious and complex fraud. Perhaps with the use of AI it may be possible to greatly enhance the ability of the SFO to identify private information and segregate it on the basis either that it is irrelevant or sensitive. In relation to the residual risk, a balance has to be struck as between privacy and the public interest in the effective prosecution of fraud.

The second objection relates to the threat of a defence ambush of the SFO at trial. Without prior warning an item of unused material is adduced in the course of a cross examination of a prosecution witness and it appears to undermine the case against the accused. Amidst trial the SFO lacks adequate time to consider it and if justified, to correct a false impression or find contextual documents amongst the unused material that dilutes the new exhibit’s significance. But this risk of unfairness can be ameliorated. A new rule of procedure could be promulgated that the defence shall not be permitted to adduce a document derived from the unused material unless it has given the SFO 14 plus days’ notice of it and provided particulars of where it was discovered.

Finally, let us stand back and consider why the keys to the warehouse model is in principle more likely to deliver a better outcome than the incumbent. Our system for trying criminal cases is adversarial. Investigators and prosecutors are not and, whilst that system endures, are never going to be impartial. The investigative mindset influenced by the plight of victims whom they encounter and the inescapable fact that managing disclosure is often dull and the least regarded task means that frankly it is idealistic and unrealistic to expect the prosecution team to be focused on disclosing material that potentially undermines its case. That team is very likely to exhibit confirmation bias (i.e. cherry picking) when examining the material. Exculpatory material may be unintentionally ignored. No amount of training is going to accomplish a paradigm shift whereby those individuals have an entirely open or neutral mind. Therefore the best people to examine the material on behalf of the accused are, unsurprisingly, the defence team. Additionally, they are better equipped to do so because they are privy to the accused’s account of events.

This article has solely concerned SFO vases. In respect of mainstream criminal cases where the quantity of unused material is likely to be much less, there may be countervailing factors which may make the case for restoration of the common law regime less compelling.

Latest Insights


May 10 2024


April 26 2024


April 22 2024