At the end of March, the Justice Department’s Inspector General published a report providing interim results in his ongoing audit of the FBI’s procedures for obtaining secret warrants under the Foreign Intelligence Surveillance Act (FISA). In collaboration with the Reiss Center on Law and Security at NYU School of Law, Just Security asked a group of four leading experts to answer specific questions about the report. We are grateful to this outstanding group whose individual responses are provided below:
Liza Goitein, Director of the Brennan Center for Justice’s Liberty and National Security Program
Andrew McCabe, former Acting Director and Deputy Director of the Federal Bureau of Investigation; CNN Legal Analyst
Mary McCord, former Acting U.S. Assistant Attorney General for National Security; Legal Director and Visiting Professor, Georgetown University Law Center’s Institute for Constitutional Advocacy and Protection
Julian Sanchez, Senior fellow at CATO Institute and a founding editor of Just Security
In December 2019, Inspector General Horowitz issued a report finding serious errors and omissions in the FBI’s warrants against Carter Page. Following the Office of Inspector General (OIG) report, Just Security published a series of articles by former government officials and civil society experts–including George Croner, Liza Goitein, Julian Sanchez, Andrew Weissmann–and the Reiss Center and Just Security convened a live panel event on the topic of FISA reform.
Meanwhile, the OIG continued to engage in a broader audit of FISA warrant applications to review the FBI’s compliance with its “Woods Procedures”—an internal process meant to ensure, in the absence of a traditional adversarial process, that there is supporting documentation for all of the factual assertions made in the application for a FISA warrant before the federal court.
In its March interim report, the OIG examined the Woods Files for a sample of 29 FISA applications, including both counterintelligence and counterterrorism investigations, selected from more than 700 applications from eight field offices over the period from October 2014 to September 2019. The OIG’s findings were sobering: four out of the 29 FISA applications were fully missing their required Woods Files; each of the remaining 25 contained “apparent errors or inadequately supported facts,” with an average of 20 errors per application. The OIG also reviewed 34 separate “accuracy review reports” conducted by the FBI and Department of Justice’s National Security Division, covering the same time period and sample of field offices. Despite the deficiencies uncovered by those agency checks, the OIG found that the FBI did not act on them to “help assess the FBI’s compliance with its Woods Procedures.” In response, the Foreign Intelligence Surveillance Court (FISC) ordered the government to provide it with further information on the 29 applications and to assess the materiality of the discovered problems.
We posed four questions to our experts, asking them to respond to at least two of the questions. The following provides their full responses, organized by topic question.
1. The OIG says it “[does] not have confidence that the FBI has executed its Woods Procedures in compliance with FBI policy, or that the process is working as it was intended to help achieve the ‘scrupulously accurate’ standard for FISA applications.” How serious are the deficiencies that the OIG identifies in this recent report? Are there mitigating factors or limits to the conclusions that can be drawn from this memo—for example, what should we make of the fact that the IG has not assessed the materiality or significance of the numerous errors it identified?
GOITEIN: The IG’s findings are unquestionably significant, and deeply concerning. The IG reviewed 29 FISA Title I applications, prepared and submitted by eight different FBI field offices, to see whether each factual assertion in the applications was supported by documentation in the associated Woods Files. This documentation requirement is the primary mechanism by which the FBI attempts to ensure “scrupulous accuracy” in its FISA applications. Yet, in four of the 29 cases, the FBI could not locate any Woods File. In every single one of the remaining 25 cases, there were multiple factual assertions in the application that were either not supported by, or were inconsistent with, the Woods File documents. The IG found an average of 20 “issues” per application reviewed, with one application containing 65 problems. According to the IG, routine audits conducted by the FBI and the Justice Department’s National Security Division over the past five years had uncovered a similar level of error.
There are no silver linings in the report. True, the IG did not attempt to determine whether the factual assertions in question were material, or whether they had evidentiary support that simply didn’t make it into the file. But the IG’s more extensive review of the Carter Page applications didn’t reveal much in the way of mitigating factors—and there is simply no reason to imagine that the Page applications were atypical. After all, the IG found no evidence whatsoever that the FBI’s investigation of Page was motivated by political bias. If anything, the IG found that the National Security Division exercised increased oversight of the Page application process.
MCCABE: It is serious any time the FBI fails to comply with its own policies, and especially serious when the policy directly implicates FISA accuracy. The intrusive nature of FISA surveillance demands that the applications for authorization be, in all material respects, absolutely accurate — which is why we need to know more than what the interim IG report tells us. Most policies are designed to ensure that an agency meets its legal and ethical obligations in a consistent and effective way. The Woods policy is no different. It requires documentation of the facts presented to the court in order to ensure the FBI meets the legal requirement of presenting a factually accurate request. It is important to note that the IG report does not find that the applications they reviewed were factually inaccurate, but simply that they contained facts not adequately documented in the Woods file. The IG’s review was extremely narrow – the review focused exclusively on comparing the FISA application to the Woods file. The IG did not review the whole case file — which may very well have contained the documents that were lacking in the Woods file. The failure to maintain a good Woods file is serious, but not nearly as serious as making misrepresentations to the court. This report does not tell us if that happened. The answer is something we still don’t know and need to know.
We also don’t know if the facts the IG found not to have adequate Woods documentation were actually material to the application. The IG report points out that standard DOJ accuracy reviews of FISA applications and their Woods files typically uncover errors, but rarely ever find errors that are considered “material.” These reviews include a robust exchange between the case agent, her supervisor, and the DOJ reviewer. They include not just the Woods file, but the entire case file and all the information known to the case agent.
Admittedly, the IG may be correct that the FBI’s comprehensive, strategic examination of these accuracy reviews “would have put the FBI on notice that the Woods Procedures were not consistently executed thoroughly and rigorously.” But that does not say as much it sounds. It just brings us full circle. Even if the FBI were on notice for such inconsistencies, we still don’t know whether that points ultimately to material errors and to misrepresentations made to courts, or instead to nonmaterial, even trivial, errors and clerical mistakes in failing to put the right documents in the right place.
Bottom line: policy compliance failures are serious, especially those involving the use of FISA. Faulty Woods files are a bad sign for FISA accuracy. But we won’t know if there were serious errors in FISA applications until someone reviews the complete case file for each request. That said, we now know better where to look and what questions to ask thanks to the IG’s efforts.
MCCORD: The deficiencies that the OIG identified in the March 2020 Management Advisory Memorandum (MAM) are significant because they suggest that the serious lapses that the OIG identified in its December 2019 report on the Carter Page FISA applications were not unique to those applications, but instead were indicative of systematic problems in the FBI’s and DOJ’s procedures for ensuring that applications made to the Foreign Intelligence Surveillance Court (FISC) are scrupulously accurate. Although the audit that led to the March report was limited to review of compliance with the FBI’s Woods procedures, the purpose of those procedures is to ensure there is a factual basis for every assertion made in a FISA application and that any pertinent information about a cooperating human source is included. Moreover, although the March report was based on a relatively small sample size of only 29 applications, those came from 8 different FBI field offices, and problems were identified in all of them.
This is a Brady moment for FISA.
By that, I’m referring to the U.S. Supreme Court’s seminal case of Brady v. Maryland, which established that prosecutors must disclose to criminal defendants material information that is exculpatory (meaning it tends to suggest the defendant is not guilty) or impeaching (meaning it tends to undermine the prosecution’s evidence). The botched 2008 trial of former senator Ted Stevens (R-Alaska), after which it was determined that trial prosecutors had intentionally concealed documents that would have helped Stevens defend himself, led the Justice Department to revamp its policies to require prosecutors to take a broad view of materiality and err on the side of more disclosure rather than less. The Department implemented this policy by issuing new guidance to prosecutors, requiring yearly mandatory training, and making a serious push to change the culture. It was a paradigm shift.
Although the OIG did not assess the materiality of the deficiencies it found in the Woods procedures—and it is possible that most were immaterial—the deficiencies identified by the OIG in both recent reports suggest that a similar paradigm shift is warranted in the FISA application process. The FBI and DOJ already are implementing process changes and training that should address some of the obvious sloppiness in complying with the very procedures designed to ensure accuracy, including in identifying material exculpatory and impeaching information that should be disclosed in any FISA application. That means both information that tends to suggest the target is not an agent of a foreign power as well as information that tends to undermine the basis on which the government is relying to establish probable cause that he or she is. The FISC is right to demand such procedural reforms in light of the OIG revelations, and to require personal accountability from agents and DOJ attorneys.
SANCHEZ: Justice Department Inspector General Michael Horowitz’s blistering report the on the error-riddled Foreign Intelligence Surveillance Act process used to wiretap former Trump campaign advisor Carter Page left a critical question unanswered: Were the problems Horowitz documented an outlier—products of aberrant recklessness or political bias—or were they symptomatic of more pervasive failures in the FISA process?
The preliminary results of Horowitz’s follow-up audit, as documented in the recently released “advisory memorandum”—strongly suggest the latter. The IG’s office reviewed, or attempted to review, the FBI’s compliance with the “Woods Procedures” designed to ensure the accuracy of factual information in applications submitted to the Foreign Intelligence Surveillance Court by looking at a sample of 29 applications for surveillance of U.S. persons submitted between October 2014 and September 2019. They did so by examining the “Woods file” that should have been generated during the review of each application, compiling documentary support for each factual representation made to the court. What they found was, to put it mildly, not encouraging.
In 4 of the 29 instances, the IG’s office “could not review original Woods Files” because the FBI was unable to locate them, and in 3 of those cases was uncertain they had ever existed. That amounts to an admission that they can’t be sure a meaningful factual review was done at all for 10 percent of the sample they audited—or at the very least, that whatever was done, was done by a reviewer who did not understand one of the most basic elements of how the review process was supposed to be executed.
Every one of the applications for which the IG was able to locate a Woods file was riddled with unsupported or inaccurate claims: “an average of about 20 issues per application reviewed, with a high of approximately 65 issues in one application and less than 5 issues in another application.”
That implies that, strictly with regards to the Woods Procedures issues, the Carter Page applications were actually significantly better than average. The appendix to the IG’s report on the Page investigation identifies eight distinct Woods failures—cases where the Woods file either failed to support or actively contradicted representations in the application—in the initial FISA application targeting Page, rising to 16 by the final renewal application.
A raw count, of course, does not in itself tell us how serious the inaccuracies were, and the IG’s memo explicitly makes no effort to pass judgment on the “materiality” of the inaccuracies they found. The Woods errors identified in the Carter Page applications included several that were arguably trivial—claims that were probably true even though not documented in the Woods file, or would be of relatively little importance even if they had been inaccurate, such as the exact dollar amount a confidential FBI source had been compensated. But, as with the Page applications, the sloppiness implied by the sheer quantity and prevalence of inaccuracies in itself belies the Bureau’s commitment to “scrupulous accuracy” in its representations to the court: If the review process is so systematically faulty, there is no reason to imagine that only insignificant errors would slip through the cracks.
Nor should these problems distract from the fact that Woods failures were not, ultimately, the most serious problems identified in the IG’s review of the Page investigation. Rather, the most egregious failures were omissions of new information, especially in the later renewal applications, that tended to undermine or contradict the FBI’s theory of the case. Woods review is designed to ensure that what’s in a FISA application is accurate; it does not establish whether critical facts the Court should be aware of have been left out entirely. Or, in the rather understated phrasing of the IG memo, the Woods Procedures “are not focused on affirming the completeness of the information in FISA applications.” Thus we still have no idea how many errors of omission mar the applications in the IG’s sample, above and beyond the Woods failures identified.
The memo also confirms that some of the institutional sources of the problems in the Page investigation were not limited to that probe. One clear trend the IG found there was that the defects in successive applications grew more serious over time. While the initial application and renewals were all flawed to varying degrees, it was only with respect to the final two renewal applications that the Department of Justice ultimately assessed the problems to be so serious that they acknowledged probable cause did not exist for the surveillance orders that ultimately issued.
One reason for this pattern, the IG report indicated, was that notwithstanding official FBI policy, renewal applications were typically scrutinized only for the accuracy of new factual representations. Claims already accepted by the court in prior applications did not receive similarly scrupulous review to see whether they were still tenable in light of new information the investigation had uncovered. Thus one of the most serious omissions in the later Page applications: the failure to document contradictions between initial reporting from former British intelligence officer Christopher Steele and later FBI interviews with Steele’s sources.
Horowitz found a similar practice reported by agents interviewed during his audit, noting: “we were told by the case agents who prepared the renewal applications that they only verified newly added statements of fact in renewal applications because they had already verified the original statements of fact when submitting the initial application.”
2. Both the FBI and the Office of the Deputy Attorney General provided replies to the memo in which they agree to both of the IG’s two recommendations. Yet overall, they indicate that mitigation steps are already being taken as part of the changes ordered in response to the IG’s initial report—in essence, that things are already moving in the right direction. What is your view of what needs to be done, and what the FBI and DOJ are already doing? What would you recommend the IG take up in his next steps in this ongoing investigation of the FISA process?
MCCABE: The steps outlined by the FBI address the concerns raised in the recent IG report by essentially raising the stakes of accountability on the current requirements. New training initiatives, a revised Woods form and the inclusion of the Woods file in the electronic case file are all good steps, but none fundamentally change the process of documenting each fact in the Woods file.
It also seems the FBI has not addressed some of the bigger issues raised in the first IG FISA report (released last December). For example, detecting and correcting errors of omission remains a systemic challenge. When a case agent collects information during the course of an investigation that might contradict facts presented to the court in an earlier FISA application, she must make an important judgement about whether that needs to be brought to the attention of her supervisor and the DOJ attorney drafting the application. These investigative developments, and the consequential judgements that follow, can happen hundreds of times across the scope of a fast paced, stressful investigation. The decisions are rarely, if ever, black and white. DOJ and the FBI need to change the process to get the attorneys more involved in assessing the investigative facts to reduce the chances of errors of omission. Simply requiring the case agent to sign an additional certification will put more stress on an already overtasked agent, but is unlikely to alter the way these factual/legal questions are being resolved.
MCCORD: The mitigation measures that the FBI and DOJ are implementing are important steps in the right direction, but as the chief judge of the FISC, U.S. District Judge James Boasberg, determined in his March 5 and April 3 Orders, more is needed. Forms and checklists are useful tools, but it takes a cultural change so that FBI agents and DOJ attorneys are socialized from the outset to be equally diligent in seeking and presenting “Brady” information as they are in seeking and presenting any evidence of probable cause. Part of this will come with more procedures and training, and part will come with pressure from the FISC. When the courts started asking federal prosecutors far more questions about their Brady obligations in criminal cases, and holding prosecutors accountable for lapses, the culture within the Department changed. Now that the OIG has brought the issues to light in the FISA realm, the same pressures that drove culture change in criminal prosecutions are likely to drive change in FISA applications.
3. What should be the role of the courts now? What about Congress?
GOITEIN: The FISA Court can no longer trust the accuracy of FISA Title I applications. Until the Department’s own audits (including follow-up IG audits) show that the problem has been resolved, the Court might need to conduct its own review of the Woods Files—or perhaps even the case files—before issuing Title I surveillance orders, at least where U.S. persons are the target. Alternatively, the Court could appoint amici to conduct this review. Either way, this review will slow the process considerably, but the alternative is for the Court to authorize surveillance that might well violate the Fourth Amendment.
Other courts also have a duty to respond. In criminal cases where the government has notified the defendant that evidence was obtained or derived from a FISA order, the court may disclose the FISA application and supporting materials to the defendant, “under appropriate security procedures and protective orders … only where such disclosure is necessary to make an accurate determination of the legality of the surveillance.” Courts have consistently found that such disclosure is unnecessary, and that they can rely on the application materials themselves. That must now change. Courts must give defendants a chance to identify the errors that they are best positioned to detect, in order to assess whether the applications were so deeply flawed as to render the surveillance unlawful.
Finally, Congress must legislate reforms. In May, the Senate is scheduled to vote on a House bill reauthorizing three provisions of FISA that expired on March 15. That legislation provides only weak civil liberties protections; the IG’s report makes clear that stronger safeguards are necessary. Among other things, the Senate should amend the House bill to strengthen the role of amici (for instance, they could be tasked with reviewing FISA Title I applications targeting U.S. persons); to bolster the requirement that criminal defendants be notified of FISA surveillance (currently, the government can evade this requirement through a practice known as “parallel construction”); and to ensure that defendants have access to FISA applications and materials so they can challenge the lawfulness of surveillance.
MCCORD: As noted, I expect the FISC to be significantly more active in questioning FBI agents and DOJ attorneys about compliance with Woods procedures and whether anything in the FBI’s case file tends to undermine probable cause. Judge Boasberg’s March Order requires the FBI declarant and DOJ attorney to attest that “to the best of [his or her] knowledge,” all information “that might reasonably call into question the accuracy of the information or the reasonableness of any FBI assessment in the application, or otherwise raise doubts about the requested findings,” be brought to light and included in the application. This type of personal accountability, certified to the court, is something that agents and DOJ attorneys can be expected to take very, very seriously. And Congress, should it choose to do so, could memorialize these and other requirements in the statute.
4. The IG’s initial findings as related to the Carter Page/Crossfire Hurricane investigation became part of a political narrative on both sides of the aisle. What will be the political ramifications of the findings described in this memo, if any?
MCCORD: There will no doubt be some who say that the FBI can’t be trusted and that FISA should be scaled back significantly. But FISA remains a critical tool to combat terrorism and other national security threats. That’s why it is important for the serious deficiencies identified by the OIG to be addressed so that the FISA court can have well-founded confidence in the accuracy of the applications brought before it.
SANCHEZ: These recent findings weigh heavily against the optimistic notion that the defects in the Carter Page investigation were some kind of extraordinary anomaly—the product of an extraordinary and rushed investigation, or subconscious political bias, or (most outlandishly) a deliberate conspiracy against Donald Trump by some sort of Deep State cabal within the FBI and Justice Department. The IG audit looked at FISA applications from eight different field offices. The period examined fell half within the Obama administration, half within the Trump administration. The cases were a mix of counterterrorism and counterintelligence cases, most presumably with no particular connection to domestic politics. And on average the sample included more Woods procedure errors than were found in even the most error-riddled of the Page applications.
While it’s doubtful this will significantly disrupt the politically convenient narrative of an anti-Trump vendetta as the fountainhead of all the FISA ills diagnosed by the inspector general, it ought to. It is further evidence that there is something badly, systematically broken in even the component of FISA—Title I—that most closely resembles the traditional criminal warrant process, and which on its face appears to have the most rigorous and multilayered oversight. And it is a strong indicator of the urgent need to continue turning over rocks.
5. Other thoughts/comments?
GOITEIN: I was particularly struck by the IG’s observations about the internal audits (“accuracy reviews”) conducted by the FBI and the National Security Division (NSD). Over the course of five years, these audits consistently revealed an intolerable level of error in FISA Title I applications: 390 errors in the 39 applications reviewed. Although the reviews found that none of the 390 errors was material—a finding the IG did not attempt to corroborate—it is hard to imagine that the cumulative effect of roughly 10 errors per application would not affect the overall soundness of the package.
And yet, the FBI made no systemic changes in response to these findings. It did not use the audits to hold individuals accountable, to revamp its trainings, to assess the efficacy of the Woods Procedures, or to examine more broadly the problem of accuracy in FISA applications. (The NSD reported that it uses the audit results in its trainings, although the IG did not follow up on this.) It is possible that case agents in FBI field offices or attorneys within the NSD took steps to correct the errors in the flagged applications; the IG did not make inquiries on this point. But at least within FBI headquarters, it appears that these audits were just a box-checking exercise—oversight for its own sake.
The FBI’s failure to make any changes to its systems in response to the consistently poor audit results raises concerns about how effective such internal audits are in other contexts. Secretive intelligence programs, in general, tend to rely heavily on internal oversight measures. Intelligence officials often point to the multiple layers of review within agencies as evidence that Americans’ civil liberties are well-protected. Clearly, however, the mere existence of audits is insufficient. Oversight will not lead to improved compliance unless it is paired with accountability. This is a key lesson to remember whenever Congress or an executive branch agency proposes additional oversight as a solution to a civil-liberties concern.
MCCORD: Just as in the Brady context in criminal cases, there are usually three possible explanations for why exculpatory or impeaching information is not provided in a FISA application. One is actual bad faith on the part of the FBI agent or DOJ attorney—a deliberate attempt to mislead the court. The OIG reports did not find such bad faith, and I expect it would be the rare explanation. Another is mistake—the agent who learns the information fails to get it into the case file or share it with the case agent. The OIG reports certainly bear out that this has happened repeatedly. And the last is failure to recognize the significance of the information. In the realm of criminal cases, a simple illustration of this might involve four eyewitnesses to a crime, three of whom identify the defendant and one who doesn’t. The prosecutor might rationalize away the significance of the non-identifying witness. Maybe that witness’s view was obstructed and the other witnesses were closer and had clear views; maybe that witness observed the defendant only for a split second and the other witnesses saw the defendant for longer or were familiar with the person; maybe that witness’s description of the entire event differed significantly from the other three witnesses and therefore seemed incredible. The prosecutor might conclude that the non-identifying witness must be mistaken and his or her information is not material. But DOJ’s policies appropriately require disclosure of the non-identifying witness’s information to the defendant. The same type of illustration can be conjured in the FISA context—some piece of information might be easily and erroneously discounted as not actually undermining probable cause. The key, as with Brady, is to take a broader view of materiality and disclose the information anyway. That means the FBI, DOJ, and the FISC all will have the tools necessary to determine whether the application should be made and whether it should be granted.
Editor’s note: with special thanks to Alexander Potcovaru for his contributions to this project.