Earlier this month, Adam Klein, the outgoing chair of the Privacy and Civil Liberties Oversight Board, took the unusual step of issuing a unilateral “Chairman’s White Paper” on oversight of the Foreign Intelligence Surveillance Act, based on PCLOB’s review of 19 FISA applications for electronic surveillance of U.S. persons in counterterrorism investigations. This is in itself notable, given how tightly restricted access to the applications underlying FISA surveillance has historically been. Until 2018, when a redacted version of the applications to monitor former Trump advisor Carter Page was declassified, the general public had never seen one. When an unprecedented deep-dive review of the Page applications by the Justice Department’s Inspector General uncovered serious deficiencies in that process, this fact took on sudden salience: Nobody could be certain whether the problems were sui generis of a larger pattern of errors and omissions. Predictably, alas, Klein’s discussion of the substantive contents of the applications PCLOB reviewed is largely redacted, but the report does offer some helpful procedural analysis and some welcome, but ultimately rather conservative, proposals for reform.

The Inspector General’s report on the Page FISA process had found fault with the Justice Department’s vaunted Woods Procedures, designed to ensure that each factual claim in an application submitted to the Foreign Intelligence Surveillance Court has documentary support in the FBI’s case file. The IG found not only mismatches between the applications and case file, or claims without documentary support but—far more troubling—material omissions of facts that weighed against the FBI’s assessment that Page had acted as an “agent of a foreign power.”  The most egregious of these occurred in the later renewal applications. Having satisfied the FISC that it had met its probable cause burden on this question in the initial application, the FBI seems to have shown little interest in revisiting whether that assessment remained tenable on the totality of the evidence as new information came in that contradicted or complicated its earlier understanding.

In response, as Klein notes, the Justice Department in mid-2020 began supplementing its “accuracy reviews” of a sample of applications with “completeness reviews”—of which it had completed 95 as of March 2021. While this represents only a fraction of the hundreds or (more often) thousands of FISA orders sought each year, it is nevertheless a nontrivial sample, and Klein observes that both the accuracy and completeness reviews absorb significant resources and person-hours. He reasonably suggests prioritizing for review applications targeting U.S. persons, and above those sought  in cases designated as “Sensitive Investigative Matters” because they raise heightened civil liberties or separation of powers concerns—because, for instance, they involve political actors, religious organizations, or the press. While this is hard to dispute as a general heuristic for allocating scarce resources, it does merit an asterisk: While much of the public discourse around FISA treats collection on U.S. person targets as the sole subject of concern, non-U.S. person FISA targets often communicate with U.S. persons, and the extent to which those communications have broader implications for domestic liberties or politics may not be readily apparent in advance of collection. One reason electronic surveillance is such a singularly intrusive tactic is that it inherently involves “searching” the communications of hundreds or thousands of (ex ante unknown) parties other than the specific target. Even if we are exclusively concerned about the privacy rights of U.S. citizens and permanent residents—itself a mistake if we understand privacy as a human right—the status of the target is at best imperfectly correlated with those equities.

Klein also notes a “gulf” between the Inspector General’s assessment of the overall accuracy of FISA applications and that of the Justice Department. An audit by the IG’s office issued in March 2020 found some form of error in every application it reviewed—and in some cases dozens—concluding that this “deficiency in the FBI’s efforts to support the factual statements in FISA applications through its Woods Procedures undermines the FBI’s ability to achieve its ‘scrupulously accurate’ standard for FISA applications.” DOJ, however, countered in a filing with the FISA Court that many of these were trivial or even merely typographical errors, and that of the tiny number of “material” errors they were prepared to concede, none undermined the overall probable cause showing. Klein proposes, again sensibly, a schema for categorizing errors identified during review, ranging from the most serious—misrepresentations or omissions of material information known to the government at the time the application was prepared—to the presumably less urgent spelling errors and typos.

Most importantly, in my view, Klein urges reevaluation of the process by which FISA renewal applications are prepared. The most serious defects identified by the IG in the Carter Page applications came in the later renewals, which omitted numerous salient facts that weakened the FBI’s case for classifying Page as a foreign agent. The Justice Department has itself acknowledged that the final two FISA applications targeting Page therefore lacked adequate predication as a result. This failure stemmed in part from DOJ’s process for reviewing renewal applications, which highlights the new information added in each iteration. While in principle renewals applications are supposed to be reviewed in their totality, in practice this appears to have led to minimal scrutiny of claims and conclusions already accepted by the FISC. Why waste time on what has already been thoroughly vetted? As Klein puts it: “The structure of renewal applications may influence the cognitive process agents and lawyers undertake in preparing them. That may encourage the drafters to rest on the facts in the original application, rather than reconsidering the probable cause assessment in light of new developments.”

We can add to Klein’s analysis that the tendency toward confirmation bias in FISA renewals is likely to be exacerbated by an important distinction between the evidentiary standards applicable to foreign intelligence surveillance, as compared with the so-called “Title III” wiretaps employed in ordinary criminal investigations. When a Title III wiretap is sought, the purpose of surveillance and the showing required before an order can issue—what we might call the “success condition” and the “threshold condition”—are reasonably tightly aligned. Barring unusual edge cases, Title III wiretaps that achieve their purpose (obtaining evidence of a crime for use in a subsequent prosecution) simultaneously reinforce their own predicates and provide additional grounds for reauthorization  (probable cause to believe that surveillance will yield evidence of a crime that has been, is being, or will be committed). A Title III order that meets its success condition, in other words, will by definition satisfy the threshold condition for its own reauthorization, assuming additional evidence is deemed necessary before commencing prosecution.  

In the case of FISA, however, the success condition and the threshold condition are not so tightly connected. The pool of U.S. persons whose communications might reasonably be deemed to contain “foreign intelligence information” under one of the five definitions delineated in 50 U.S.C. §1801(e) is almost certainly substantially larger than than the pool of U.S. persons knowingly engaged in clandestine intelligence activities at the behest of a foreign power. A communication in which an American is probed for information by members of a foreign clandestine service, for example, could very well provide information that “relates to” or is “necessary to” protect against “clandestine intelligence activities by an intelligence service or network,” or to the conduct of U.S. foreign affairs regardless of whether the American is a knowing agent or an unwitting asset.  The Inspector General’s findings in the Page case suggest that, having once persuaded the FISC of a target’s foreign agent status, the focus in renewals shifts to the question of productiveness: whether surveillance has generated, and continued surveillance is likely to generate, foreign intelligence information. But this is, of course, a different question from that of whether the initial assessment that the target is a foreign agent has been validated.

This difference may contribute to inadequate scrutiny of renewal applications by investigators and attorneys accustomed to the criminal investigative process. If a Title III wiretap is productive—yielding evidence of a crime—that productiveness inherently vindicates the initial showing that there was probable cause to believe such evidence would be obtained. Failure to adequately appreciate that the same tight nexus between “threshold” and “success” conditions need not exist in FISA surveillance may be one cause of insufficient attentiveness to new information undermining the “agent of a foreign power” determination.

While Klein’s proposals are fine as far as they go, they represent relatively modest procedural tweaks that do not, in my view, get at the root cause of dysfunction in the FISA process: the fact that “FISA applications are not tested in an adversarial process, and FISA surveillance is classified.”  FISA intercepts can be used as evidence in court—subject to the Classified Information Procedures Act, which seeks to balance defendants’ due process rights against national security interests—but this is not their main purpose, and in practice it is vanishingly rare.  The overwhelming majority of FISA targets never learn that they, or the people they communicate with, have been wiretapped.  This eliminates a critical mechanism of accountability: The target, after all, is far better situated than any DOJ reviewer to identify falsehoods or missing context. The elaborate multilayered system of review FISA applications undergo is an imperfect attempt to compensate for the absence of this mechanism—and indeed, it seems plausible that accuracy gains from requiring multiple reviewers are offset by the diffusion of responsibility this entails.   If one reviewer misses a problem, many other eyes will have an opportunity to catch it, and if everyone misses it, then (excepting really egregious misconduct) it’s hard to fault anyone in particular.

The best remedy here may also be the most straightforward: End the presumption that FISA surveillance, at least in the case of U.S. person targets, will remain permanently covert.

Permanent secrecy has been baked into FISA since its inception, and the special exigencies of foreign intelligence collection surely justify a greater degree of secrecy than we countenance in criminal investigations, where targets typically must be notified within 90 days of the termination of surveillance.  But even giving due weight to those considerations, a categorical rule of permanent secrecy in every FISA case seems impossible to justify.

Ordinarily, notice to the target of a search is constitutionally required—an element of the Fourth Amendment “reasonableness” of a search—even though it may be delayed when advance notice would frustrate the purpose of the search.  In the seminal case Berger v. New York, the Supreme Court invalidated a New York State wiretapping statute in part because it failed to adequately provide for notice:

Finally, the statute’s procedure, necessarily because its success depends on secrecy, has no requirement for notice as do conventional warrants, nor does it overcome this defect by requiring some showing of special facts. On the contrary, it permits unconsented entry without any showing of exigent circumstances. Such a showing of exigency, in order to avoid notice, would appear more important in eavesdropping, with its inherent dangers, than that required when conventional procedures of search and seizure are utilized. Nor does the statute provide for a return on the warrant thereby leaving full discretion in the officer as to the use of seized conversations of innocent as well as guilty parties. In short, the statute’s blanket grant of permission to eavesdrop is without adequate judicial supervision or protective procedures.

In United States v. Freitas, the Ninth Circuit noted that, following Berger, “the absence of any notice requirement in [a] warrant casts strong doubt on its constitutional adequacy.”  The warrant at issue in that case was found “constitutionally defective in failing to provide explicitly for notice within a reasonable, but short, time subsequent to the surreptitious entry” on the grounds that “surreptitious searches and seizures of intangibles strike at the very heart of the interests protected by the Fourth Amendment.”

There is ample reason to suppose that “exigency” would, in the intelligence context, often justify significantly longer delays than Title III permits.  Here, after all, the government is concerned not only with tipping off an individual target, but with exposing to foreign adversaries the contours of intelligence collection efforts that may span years.  FISA, however, does away with the notice requirement categorically, without any need for a particularized showing that notice of a wiretap would incur harms.  And it does so even when, as in the Carter Page case, the government’s initial assessment that a target is acting as a foreign agent can’t be sustained.

The Page case, of course, provides us with at least one instance where it was evidently possible to publicly disclose not just the fact of surveillance, but substantial detail about its basis, without apparent injury to national security.  In the rare cases where FISA evidence is used in a criminal prosecution, defendants similarly receive notice, subject to the constraints of the Classified Information Procedures Act.  But if it is at least sometimes possible to provide notice without imperiling national security, FISA’s categorical presumption of permanent secrecy seems impossible to justify.  Even giving maximum deference to the special needs inherent to intelligence collection, it cannot be that a constitutional notice requirement is overcome simply because the government at one point believed a target to be a foreign agent, regardless of whether that belief is borne out.

FISA should instead require a particularized “showing of exigency to avoid notice” at the termination of surveillance, at least in the case of U.S. person targets.  Even if there are indeed compelling grounds for extended delays of notice in most such cases, eliminating the universal presumption creates at least the prospect of meaningful accountability for improper targeting of Americans.

One final feature of the FISA process—a matter of longstanding practice rather than statutory structure—deserves reevaluation. The FISA Court is frequently defended against charges that it functions as a “rubber stamp” on the grounds that its high rate of approved applications does not capture the extended dialectical process that occurs between FISC staff and DOJ attorneys.  Proposed applications are often modified—or withdrawn—in response to feedback from the court, rather than being submitted and rejected.  The FISC, in other words, is more scrupulous and demanding than the official approval statistics imply.  

This process, however, may itself have undesirable consequences.  Like any court, the FISC relies on a body of precedent to guide its rulings.  Uniquely, however, this body of precedent consists primarily of the FISC’s own classified opinions interpreting the FISA statute and the requirements of the Fourth Amendment in the foreign intelligence context.

Normally, precedent functions to set boundaries on government conduct by establishing exemplars of both what is permissible and what is forbidden.  Under this set of facts, a government search comported with the requirements of the Fourth Amendment; under that set of circumstances, it did not.   Many of the cases as the heart of our Fourth Amendment jurisprudence are, in essence, instances of a court telling the government “no.”  (Typically, of course, these are the result of a challenge raised by the subject of a search—a challenge that the secrecy default denies most FISA targets the opportunity to raise.) These establish benchmarks to which future courts can refer in evaluating new cases: If past government conduct was held incompatible with the Fourth Amendment, a similar result should obtain when similar facts recur.

The informal FISA dialectic, however—especially in the absence of adversarial testing after the fact—biases the paper trail, so that what survives as precedent is disproportionately a record of “yesses.”  When the FISC approves an instance of electronic surveillance, that approval is preserved for the reference of later FISC judges.  Yet when the court informally rejects a draft application—because its scope is too broad, or its proposed minimization procedures inadequate, or its probable cause showing thin—a comparable benchmark may not be established.  A new FISC judge confronting an application may have no easy way of knowing that substantially similar applications were in the past rejected at a preliminary stage, while approvals under similar facts remain etched in the record.

Given the natural variance in judicial attitudes to close cases—there will always be marginal fact patterns where some judges would say “yes” while others would say “no”—the FISA dialectic risks setting up a ratchet effect over time, even if we stipulate that the Court’s procedures are well designed to prevent deliberate judge-shopping.  A dozen FISC judges may informally say “no,” but it is the one who eventually says “yes” who furnishes the government with a citation for future use.

While this informal dialectic doubtless has many advantages, both for the FISC and the Justice Department, over time it is likely to have a cumulative distorting effect on the FISC’s determinations. This is particularly troubling when we consider the FISC’s evolving role, which has grown beyond the largely ministerial approval of specific targets, and now routinely includes the evaluation of programmatic surveillance. While it may be less convenient in the short term, it is likely improve the quality of FISC deliberation if the court gets in the habit of putting more of its “nos” on the record.

Such fundamental changes to the FISA process are, needless to say, more difficult to implement and more prone to meet resistance than Klein’s more incrementalist proposals.  But if the problems with the current FISA process are indeed structural, as they appear to me to be, then they will only be adequately addressed by structural reform.

Image: Spencer Platt/Getty