By Sabina Mariella and Matthew L. Schwartz
The “total institutional failure” of the Department of Justice’s handling of privileged material in the ongoing prosecution of former Goldman Sachs banker Roger Ng highlights the serious issues with the use of “filter” or “taint” teams, Sabina Mariella and Matthew L. Schwartz at Boies Schiller Flexner argue.
The DOJ uses taint teams to review potentially privileged material and determine what may be shared with the team investigating or prosecuting a case, and what must be withheld as privileged. But the practice, which essentially makes prosecutors the sole arbiter of whether communications are privileged, poses serious risks for both the DOJ and private parties. In the last year alone, the use of filter teams has resulted in at least one mistrial in a high-profile prosecution and substantially impaired the defence in another.
What is a filter, or “taint,” team?
During investigations, law enforcement authorities often come into possession of large amounts of potentially privileged material. Especially in white-collar investigations where it is not immediately clear whether documents are relevant or not, the government frequently takes a seize-first, ask-questions-later approach to evidence-gathering. For example, when the government raids an office, it often carts away (or images) all of the business’s servers and other computers for later review. When the government obtains a search warrant to seize a person’s e-mail or cloud account, internet service providers routinely turn over the entire contents of the account.
Possessing this much potentially privileged information is risky for the government. If prosecutors improperly review privileged communications, that exposure may taint the entire investigation and make prosecutions difficult or even impossible. To show that a prosecution has not been tainted by exposure to privileged information, prosecutors have to be able to demonstrate that none of their subsequent investigative activities or charging decisions were improperly based on privileged information. Enter filter, or “taint,” teams.
The filter team consists of prosecutors and/or agents who review seized material to determine whether it is protected by privilege. The filter team is separate from the team assigned to investigate or prosecute the matter, often called the “case” or “trial” team, although both teams are part of the same Department of Justice. Further, the case team and filter team are not truly walled off from one another. The case team often provides information about its investigation to the filter team, or answers the filter team’s questions. Once the filter team completes its review, it provides the case team with documents it deems non-privileged.
Aside from these definitional principles, there are no real standards to guide how filter teams conduct their review. The Justice Manual (formerly known as the US Attorney’s Manual) sets forth only the minimal requirements that the filter team not consist of attorneys or agents on the case team, and that the review procedures should be “discussed prior to approval of any warrant.” But the manual does not require court approval of privilege determinations or review protocols, nor does it call for allowing privilege-holders to review seized documents to provide their input, or to object to a determination of the filter team.
The problem with having the government make privilege calls
The use of filter teams, and the lack of standards guiding them, has proven to be seriously problematic. To start, whether a document is privileged is often a fact-intensive inquiry into the relationship between individuals and the purpose of a communication. Although the filter team is separate from the case team, it is still part of the prosecuting agency and may have an inherent bias towards seeing the prosecution succeed – or just an inherent bias in favour of doctrines like the crime-fraud exception.
A 2019 opinion from the Fourth Circuit recognised these risks. In that case, the government executed a search warrant on a law firm. The law firm sought to enjoin the filter team from inspecting privileged materials and asked that a magistrate judge or special master conduct the privilege review instead. The Fourth Circuit agreed, concluding that using unsupervised filter teams to make fact-based privilege determinations was an impermissible delegation of judicial functions to the executive branch. In reaching this conclusion, the Fourth Circuit raised numerous issues with filter teams, including the possibility that an unsupervised filter team could make errors and then transmit privileged materials to the case team and that the filter team may have a more restrictive view of the privilege than the privilege-holder. But that decision arose in the unusual context of a search of a law firm, and courts routinely permit filter teams in other situations even as the practice has become subject to increasing criticism.
These problems have been apparent in the ongoing criminal trial of former Goldman Sachs executive Roger Ng in the Eastern District of New York. Ng was indicted for his alleged role in one of the most high-profile foreign corruption investigations of recent years — the alleged looting of the Malaysian sovereign development fund known as 1MDB. At around 11:30 pm on the first day of trial, afterthe parties had given opening statements and the prosecution had begun examining its first witness, the prosecution notified the defence that it was producing 121,668 pages of discovery obtained through search warrants executed on the key cooperating witness’s email accounts and laptop. A filter team had reviewed the documents but had only recently released them to the case team. The prosecution also informed the defence that the production might contain statements of the defendant himself, meaning that they should have been produced much earlier.
The case team explained to the court that it had intended to produce the documents prior to opening statements, but as it was preparing the production, it identified privileged communications between the witness and his own criminal defence attorney. That is, the filter team had permitted the most basic form of attorney-client communications – communications between the witness and his lawyer about the very subject matter of the investigation – to be produced to the case team in bulk. To its credit, once the case team spotted the error, it asked the filter team to review the documents again, which caused the extensive delay.
About a week later, however, in the middle of the night and during the cooperating witness’s direct examination, the case team informed the court that the filter team had also over-withheld, and failed to release to the case team 15,000 non-privileged documents related to the witness – documents that the defence desperately wanted to have for cross-examination. The case team pointed the finger at the filter team, stating that the filter team had incorrectly informed the case team that it had completed its review and production of the witness’s documents. The government itself called this an “inexcusable error” and a “total institutional failure.”
This disconnect between two teams within the same government agency illustrates one of several problems with filter teams, which are apparently not being adequately managed and monitored.
The lack of quality control and communication that caused issues during Ng’s trial is compounded by the lack of clarity as to whether and when a privilege-holder has the right to challenge a filter team’s privilege determinations. This issue, too, reared its head during the Ng trial. Because of the filter team’s multiple late productions to the case team, in an effort to quickly make productions to the defence, the case team produced documents from the witness’s devices that it contended might be covered by the marital communications privilege. But apparently no one had informed the witness’s wife that her marital communications had not only been produced, but that they also might be introduced as evidence at a public trial. Thus, in the middle of the witness’s direct examination, his wife’s counsel wrote a letter to the court explaining that she had just learned about the multiple productions containing her marital communications. She was never afforded the chance to object, and it is still not clear how the case team got access to these potentially privileged communications.
At a status conference, the prosecution and defence sparred over whose obligation it was to notify the wife. The prosecution argued that if the defence wanted to use the documents, it had to notify potential privilege-holders and litigate any objections. The defence contended that it was the filter team’s duty to notify potential privilege-holders, and that once the case team produced documents to the defence, it was free to use them at trial.
Ng’s trial is not the first high-profile trial in which the use of a filter team has interfered with a defendant’s ability to craft a defence. This past summer, Michael Avenatti secured a mistrial in his fraud trial in California after testimony revealed that the government possessed potentially exculpatory accounting data that had not been produced to the defence. Unbeknownst to the case team, the filter team – which in that case was tasked with determining both responsiveness and privilege – had a copy of the accounting data but failed to produce it to the case team. The court found that failure to be a Bradyviolation, and that Avenatti was denied the opportunity to formulate a defence by not having the data prior to trial. The problem, as in Ng, was a failure of communication between the case team and the filter team. Yet it is far from clear that more communication is desirable, since the whole point of the filter team is that it be separate from the case team.
Fixing filter teams
The government’s use of filter teams clearly raises a host of risks and concerns. If nothing else, the Department of Justice needs to promulgate clear guidance that both informs how filter teams do their work, and that gives privilege-holders the ability to review or litigate anything beyond the most conclusory privilege calls.
First, filter protocols – whether set out in a search warrant or under DOJ policy – should provide privilege-holders a clear right to review all material deemed non-privileged before it is disclosed to a case team. This right should be afforded to privilege-holders as early as possible, and not on the eve of trial, to avoid delay and the sorts of disruptions that have marred Ng’s trial.
Second, while exceptions might need to be made in investigations that remain covert and therefore in which the privilege-holder cannot be part of the review process, the filter team should limit itself to releasing to a case team only the most clearly non-privileged material, such as communications that do not involve a lawyer at all. Anything approaching a nuanced decision should await the input of the privilege-holder or, if absolutely necessary, be made by a court. Likewise, where the DOJ and the privilege-holder disagree about whether material is privileged, the DOJ should not get to make the decision – it should be reserved for a judge or special master.
Third, and relatedly, DOJ policy should require close monitoring of the work and progress of filter teams by supervisory personnel. It is utterly unacceptable for a case team to blame a failure on the filter team, as if the filter team is part of some other organisation. The DOJ as a whole, and the case team in particular, must bear responsibility for breakdowns in communication that infringe on a criminal defendant’s rights or needlessly violate privilege.
While implementing these suggestions is both sorely needed and relatively easy to do, we are not particularly hopeful that the DOJ will relinquish its unilateral role in making privilege determinations. It will claim that privilege-holders will take too long and consume too many resources, and use the review process to thwart investigations. While interacting with privilege-holders will likely take additional time, in most cases that will be time well spent that will not prejudice the government’s investigation. In any event, it is incumbent on the government, which is charged with administering justice, to get it right. In the meanwhile, defence counsel should request as much specificity and clarity about filter team protocols as early as possible, and should promptly raise any issues with the Department of Justice, and where necessary, the courts.