Table of Contents

I. Introduction
II. Background: Courts and Congress Repeatedly Attempt to Fix Unconstitutional Gag Orders that Accompany NSLs.
III. What We Learned: FBI Data Shows That the Termination Procedures Are Wholly Ineffective at Ending De Facto Permanent Gag orders. 
IV. Why All the Silence? The Majority of NSL Recipients Don’t Publish After the FBI Gives Them Permission.
V. Recommendations    
VI. Notes

I. Introduction

This report details how, despite repeated efforts by Congress and federal courts, the FBI’s controversial national security letter (“NSL”) authority remains unconstitutional by subjecting recipients to indefinite prior restraints. NSLs combine “a ban on speech and a shroud of secrecy” in ways that are “antithetical to democratic concepts and do not fit comfortably with the fundamental rights guaranteed American citizens.”1 Over the last fifteen years, courts have held successive iterations of the NSL statutes unconstitutional, with Congress stepping in each time to fix the most glaring deficiencies. Most recently, in 2015, Congress ordered the FBI to develop procedures for reviewing and terminating NSL nondisclosure orders (“Termination Procedures”).

Using the Freedom of Information Act (“FOIA”), EFF sought more information on how the FBI’s Termination Procedures functioned in practice. In 2019, EFF prevailed in a FOIA lawsuit and obtained records that reveal the FBI’s activities pursuant to the Termination Procedures during the first 20 months of their operation.

We learned that when left to its own discretion, the FBI overwhelmingly favors maintaining gag orders of unlimited duration. Our findings suggest that the FBI’s Termination Procedures—both as written and in practice—neither carry out Congress’ stated purpose of reducing gag orders nor bear the weight placed on them by courts. They do not meaningfully reduce the large numbers of de facto permanent NSL gag orders, failing First Amendment scrutiny. They also fall short of adequately safeguarding recipients’ First Amendment rights. And as the records and data EFF obtained in its FOIA suit show, the FBI is unlikely to make progress in ending those gags without further direction by Congress or the courts.

Accordingly, this report concludes with recommendations for the legislative and judicial branches to remedy continuing problems with NSLs.

II. Background: Courts and Congress Repeatedly Attempt to Fix Unconstitutional Gag Orders that Accompany NSLs.

Summary

  • NSL statutes authorize the FBI to demand customer information from a range of businesses, including Internet companies, banks, and consumer reporting companies. They also authorize the FBI to impose a nondisclosure order or “gag” to prevent these businesses from speaking about the NSLs they receive.
  • The government’s use of NSLs has grown significantly since the passage of the Patriot Act in 2001.
  • Congress has amended the NSL statutes repeatedly in light of their controversy and unconstitutionality. The USA FREEDOM Act of 2015 requires the FBI to issue procedures to review outstanding NSL gag orders and terminate them when they no longer serve their intended purpose.
  • Because courts have found these Termination Procedures to be legally significant, EFF filed a Freedom of Information Act lawsuit to learn how the FBI has been administering them.

A. National Security Letter statutes authorize the FBI to collect information held by businesses and demand their silence.

National Security Letter statutes2 allow the FBI and other authorized agencies to request information relevant to national security investigations from a wide range of businesses and individuals, including electronic communications service providers, banks, consumer reporting agencies, and other financial institutions. The statutes also allow the government to issue nondisclosure or “gag” orders that prohibit recipients from disclosing any information about the request—including the simple fact that the recipient received an NSL. No court reviews or issues NSL gag orders in advance; instead, they are based on a certification by an FBI official that “that the absence of a prohibition of disclosure” may result in one of several enumerated harms.3

In practice, the government unilaterally imposes an NSL nondisclosure order in nearly every case—97 percent of the time by one estimate.4 Although the first NSL statute was enacted in 1986,5 the use and scope of NSLs expanded dramatically with the passage of the USA PATRIOT Act in 2001.6 Since that year, the government has issued hundreds of thousands of NSLs. Although the pace has slowed somewhat in recent years, the government continues to issue up to 20,000 NSLs annually, accompanied by nearly 40,000 requests for information.7

NSLs have also long been a matter of significant public controversy. In a series of reports issued between 2007 and 2014, the DOJ Office of the Inspector General (OIG) documented the agency’s systematic and extensive misuse of NSLs. The OIG concluded that, left to itself to ensure that legal limits were respected, “the FBI used NSLs in violation of applicable NSL statutes, Attorney General Guidelines, and internal FBI policies.”8

The NSL statutes’ speech restrictions prompted at least three federal courts to hold that the nondisclosure orders authorized by prior versions of the NSL statute violated the First Amendment.9 As a federal district court explained in 2013, the gag orders that accompany NSLs are “especially problematic in light of the active, continuing public debate over NSLs, which has spawned a series of Congressional hearings, academic commentary, and press coverage.”10 NSL nondisclosure orders shape discourse by preventing recipients themselves from participating in the public dialogue on governmental surveillance and from making true and complete reports to customers, investors, and legislators. Internet companies have employed a number of methods to bring transparency to the government’s use of NSLs, including so-called warrant canaries,11 transparency reports, and lawsuits to challenge the enforcement of NSL gag orders.12 During the course of constitutional challenges to NSLs, Congress periodically amended the NSL statutes to address some of the most egregious problems.13

B. Congress addresses problems with NSLs by requiring FBI to periodically review NSL gag orders.

Prior to 2015, the NSL statutes had no requirement that the FBI review nondisclosure orders after any length of time to determine whether the Bureau still believed recipients should remain gagged. As a result, most NSL gag orders were de facto permanent. In response to public pressure and adverse court rulings, Congress enacted the USA FREEDOM Act of 2015.14 USA FREEDOM included reforms designed to address constitutional defects in the NSL statutes,15 but the constitutionality of the amended statutes remains the subject of litigation.16

USA FREEDOM required the FBI to “review at appropriate intervals” NSL nondisclosure orders to determine whether the underlying facts still supported preventing the NSL recipient from disclosing the nature or existence of the NSL, and to “[terminate] such a nondisclosure requirement if the facts no longer support nondisclosure.”17

The FBI then developed procedures, effective in February 2016, that “govern the review of the nondisclosure requirement in NSLs and termination of the requirement when the facts no longer support nondisclosure.”18 The NSL Termination Procedures require the FBI to review nondisclosure orders on (at most) two occasions: (1) the three-year anniversary of the initiation of a full investigation involving an NSL and/or (2) the close of such an investigation.19

The procedures instruct FBI agents to release NSL recipients under review from those nondisclosure orders “unless the FBI determines that one of the statutory standards for nondisclosure is satisfied.”20 That is, agents must once again assess whether permitting an NSL recipient to discuss the fact of receiving an NSL “may result in (i) a danger to the national security of the United States; (ii) interference with a criminal, counterterrorism, or counterintelligence investigation; (iii) interference with diplomatic relations; or (iv) danger to the life or physical safety of any person.”21

The NSL Termination Procedures also require FBI officials to individually review each NSL issued in an investigation “to determine if the facts no longer support nondisclosure under the statutory standard for imposing a nondisclosure requirement when an NSL is issued.”22 When an FBI official determines that “nondisclosure of an NSL is no longer necessary,” i.e., the statutory factors permitting the issuance of a nondisclosure order are no longer present, the FBI provides written notice to the NSL recipient that the nondisclosure provision has been terminated (“NSL termination letters”).23

As a result of the procedures, the FBI has determined that the nondisclosure requirements accompanying a number of NSLs should no longer remain in effect. Released from the FBI’s nondisclosure requirements, several recipients, including Google, Microsoft, Twitter, Yahoo, Facebook, Cloudflare, and CREDO Mobile have published the underlying NSLs as well as the accompanying termination letter.24 But as described more fully below, many other NSL recipients have not gone public even after receiving similar termination letters.

C. A federal appeals court upholds constitutionality of NSL statutes based in part on the FBI’s Termination Procedures.

The experiences of two NSL recipients EFF represents in constitutional challenges to the NSL statutes and documents disclosed by the FBI show that the Termination Procedures do not solve the unconstitutionality of the NSL statutes. EFF represents service providers CREDO Mobile and Cloudflare, who have been subject to NSL nondisclosure orders since 2011 and 2013, respectively, and have brought lawsuits challenging these NSLs on constitutional grounds.

In defending the constitutionality of the speech restrictions imposed by NSL statutes, the government argued that the NSL Termination Procedures remedy constitutional defects in the NSL statutes, and in 2017, the U.S. Court of Appeals for the Ninth Circuit agreed in part.25 The Ninth Circuit acknowledged that even the amendments to the NSL statute in USA FREEDOM raised significant concerns that NSL gag orders of indefinite duration would not be narrowly tailored under the First Amendment, but it found that the NSL Termination Procedures “largely address this concern.”26

However, the FBI’s application of its Termination Procedures to CREDO and Cloudflare expose the unconstitutional gaps in these procedures and the NSL statute. CREDO has been subject to an NSL gag order since 2011, which appears to be permanent. This is because the FBI has closed the underlying case associated with the NSL it received in 2011.27 By their own terms, the FBI’s Termination Procedures do not require the FBI to ever reconsider a nondisclosure order once an underlying investigation closes. Thus, the prohibition limiting what CREDO could say about the NSL—including not being able to notify the subscriber whose information the FBI requested—remains in effect. And neither the Termination Procedures nor the NSL statutes require the FBI to ever again reconsider the gag it has imposed on CREDO.28 Meanwhile, the nondisclosure order the FBI entered against Cloudflare in 2013 remains in place indefinitely. This is because the NSL Termination Procedures only require subsequent review of the gag order at the close of the FBI’s investigation. Cloudflare thus is in the same position as thousands of other NSL recipients, and its ability to fully exercise its First Amendment rights remains entirely dependent on the FBI’s discretion.

At least one federal court has noted that the Termination Procedures create “loopholes” permitting indefinite gag orders against CREDO, Cloudflare, and other NSL recipients. Among the procedure’s gaps, the FBI conducts no further review beyond the two circumstances outlined in the procedures, “meaning that where a nondisclosure provision is justified at the close of an investigation, it could remain in place indefinitely thereafter.”29 Also, the procedures only apply to “investigations that close and/or reach their three-year anniversary date on or after the effective date of these procedures,” and, as a result, “a large swath of NSL nondisclosure provisions [that predate the procedures] may never be reviewed and could remain unlimited in duration.”30 And in the case of “long-running investigations, there could be an extended period of time – indefinite for unsolved cases-between the third-year anniversary and the close date.”31 Another provider has highlighted these and other problems with indefinite NSL gag orders as part of a challenge pending before the Ninth Circuit.32

The resulting gaps created by the Termination Procedures leave thousands of NSL recipients perpetually gagged. But, as explained below, the FBI’s implementation of its own Termination Procedures results in yet more problems. That is, even when the procedures do result in the review of NSL gag orders, the FBI is overwhelmingly concluding to maintain gag orders rather than dissolve them.

D. EFF files FOIA lawsuit to determine whether the Termination Procedures are working to end indefinite NSL gag orders.

Given the weight the Ninth Circuit placed on the NSL Termination Procedures in upholding the NSL statutes, their nature and efficacy has a measure of constitutional significance. Thus, EFF sought to know more about how the FBI was implementing the procedures and whether those efforts resulted in a large reduction of the number of outstanding NSL gag orders.

In September 2016, roughly nine months after the Termination Procedures became effective, EFF submitted a FOIA request to the FBI seeking documents that would reflect the Bureau’s implementation and operation of the procedures. The request specifically asked for documents that would show how the FBI was making decisions under the Termination Procedures. Those documents would include records showing how many NSLs agents had reviewed, the decisions that resulted from those reviews, and any notices sent to NSL recipients once the FBI determined to terminate a gag order. The request also asked for any other guidance or policies created as part of the procedures.

The FBI responded that it had no responsive records, and EFF filed suit in June 2017. The FBI later began searching for records, subsequently locating all the Termination Letters it had issued to third parties informing them that they were no longer subject to NSL nondisclosure orders. The FBI also began searching for records that would show how many times it had reviewed NSL gag orders and whether agents determined they could release recipients from those orders or not.

Although the FBI provided EFF with copies of roughly 750 termination letters it sent during the time period subject to the FOIA request, it completely redacted the names of the recipients. It should be emphasized that the very function of the Termination Letters was to inform these recipients they could identify themselves publicly. Yet the FBI claimed that identifying these recipients in response to EFF’s FOIA lawsuit would disclose FBI investigative techniques. The FBI claimed that disclosing the names of parties released from NSL gag orders would, in the aggregate, reveal the Bureau’s trends about how frequently it issues NSLs and how often it sends them to particular parties.33

The district court held that the FBI’s arguments were “dubious” for several reasons: First, the termination letters represented a tiny fraction of NSL recipients, as the Bureau issued roughly 750 letters in comparison to more than 37,000 NSLs in a two year period.34 Second, because the Termination Procedures delay review of NSL gag orders until three-year anniversary of the underlying investigation or the closing of that investigation, the termination letters only reveal past decisions by FBI agents that do not forecast current or future activities.35 Third, because many NSL recipients have made public that the FBI lifted gag orders it previously imposed—and any such recipient could do so at any time—much of this information is already publicly available.36

The FBI eventually complied with the order and disclosed the names of the recipients of the roughly 750 termination letters. And as discussed in the next section, the Bureau also provided data to EFF that shed light on how few NSLs it was reviewing under the Termination Procedures and how agents decided to lift only a sliver of the gag orders they reviewed.

III. What We Learned: FBI Data Shows That the Termination Procedures Are Wholly Ineffective at Ending De Facto Permanent Gag orders.

Summary

  • Records obtained by EFF show that the FBI’s Termination Procedures only result in the review of a small fraction of NSLs issued.
  • Even when the FBI does reviews an NSL nondisclosure order, it lifts the gag less than 7 percent of the time.
  • EFF’s research suggests that the Termination Procedures will not result in an overall reduction in the number of permanent NSL gag orders.
  • The FBI has issued roughly 500,000 NSLs issued since 2001, but the Termination Procedures have resulted in ending about 800 gag orders.

A. The FBI lifts less than seven percent of the NSL gag orders it reviews, which are themselves only a tiny fraction of total gag orders that remain in effect.

The FBI’s Termination Procedures, both as written and in practice, fail to end the de facto permanent gag orders imposed by NSLs. They also fall short of adequately safeguarding recipients’ First Amendment rights. Data provided by the FBI show that the Bureau lifted only 6.4 percent of the gag orders it reviewed between February 2016 and September 2017.37 During that time period, the FBI reviewed 11,874 NSLs under the Termination procedures. Of that total, FBI agents determined in only 760 instances that the gag orders could be removed. Correspondingly, FBI agents concluded that 11,114 NSL gag orders should remain in place, upholding gag orders 93.6% of the time.38

FBI Termination Procedures in Operation (February 2016 – September 2017)39
Total number of NSL gag orders reviewed 11,874
Nondisclosure orders lifted 670
Nondisclosure orders maintained 11,114

 

Although the data obtained in our lawsuit doesn’t include more recent activity under the Termination Procedures, the fact that FBI agents determined that less than 7 percent of NSL gag orders could be lifted in the time period subject to EFF’s FOIA request demonstrates that the Bureau is failing to lift gag orders in the vast majority of NSLs it issues. The extremely low rate shows that the procedures do not fulfill their intended purpose. And, as explained below, the FBI’s low rate of terminating NSL gag orders shows that the procedures do not remedy the NSL statutes’ unconstitutionality. Left to its own discretion, the FBI overwhelmingly favors maintaining gag orders of unlimited duration. This is unsurprising. Given that FBI personnel issue gag orders along with NSLs all or nearly all of the time, the released data confirms that those same personnel are overwhelmingly predisposed to maintaining gag orders at nearly the same rate they issued them.

The FBI’s low rate of releasing NSL recipients from gag orders is even more problematic in light of the fact that under the Termination Procedures, agents may never have to revisit their decisions, resulting in de facto permanent gag orders. As described above, the Termination Procedures only require agents to review NSL gag orders at the close of the investigation or at the investigation’s three-year anniversary. Although the data disclosed to EFF does not indicate which prong of the Termination Procedures triggered the review of the 11,874 gag orders between February 2016 and September 2017, the procedures themselves only allow for a subsequent review in narrow situations: when an investigation using an NSL lasts for more than three years and is closed at a later date.

As a court critical of the procedures recognized, once the FBI reviews a gag order at the close of an investigation, nothing in the procedures requires agents to revisit a decision should they decide to maintain a gag order.40 And there may be large gaps between when agents review an NSL gag order at the three-year anniversary and the close of the investigation (if the FBI ever does close the investigation), which also results in unbounded gag orders.

Although the Termination Procedures do not provide the only mechanism to release NSL recipients from gag orders, they are the only legally required mechanism for programmatic review and termination. The other primary method of ending gag orders is when recipients bring legal challenges or notify the FBI that it must obtain a court order justifying the ongoing nondisclosure. This method places the burden on recipients to initiate judicial review, and in most cases, to do so repeatedly until the court determines that a gag order is no longer necessary. As such, it is unlikely to lead to the termination of NSL gag orders on a large scale.41

By EFF’s estimate, the combined methods of judicial review and Termination Procedures has resulted in bringing an end to roughly 800 NSL gag orders. This pales in comparison to the more than 500,000 NSLs the FBI has issued since 2001.

B. The slow rate of FBI review of NSL nondisclosure orders compounds the problem because the Bureau will be unable to review all the NSLs it issues in a timely manner.

The data EFF received from the FBI highlights an additional problem with the Termination Procedures: the FBI’s slow rate of review of NSL gag orders in the first instance makes it all but impossible for the Bureau to meaningfully review and end NSL gag orders. In short, the FBI is issuing NSLs and their accompanying gag orders at a much faster clip than it is reviewing them under the Termination Procedures. Absent clearer direction from Congress and courts (likely requiring additional resources), this trend is likely to result in an ever-growing backlog of gagged NSL recipients.

Since 2013, the FBI has issued approximately 12,000 NSLs each year, nearly all of which contained nondisclosure orders.42 The data the FBI produced to EFF showed that in 20 months – between February 2016 and September 2017 – the Bureau reviewed only 11,874 NSLs under the Termination Procedures, or roughly 594 per month . Although the limited time period of the data the FBI provided to EFF make it difficult to extrapolate larger trends, the data raise troubling questions about whether the Bureau has the capacity to meaningfully review NSL gag orders in a timely manner.

For example, even using the lowest total number of NSLs issued in a recent year as a baseline—10,235 in 2018—the FBI sends out an average of more than 850 NSLs a month. Assuming a regular rate of review based on the data obtained by EFF, that means the FBI issues nearly 260 more NSLs per month than it reviews (850 issued and 594 reviewed). In other words, the FBI likely issues far greater numbers of NSLs than it can review under the Termination Procedures and this gap will continue to grow even larger, even if the rate of NSL issuance stays unchanged.

The low review rate under the Termination Procedures also raises questions about whether the FBI has the personnel and financial resources to tackle the mountain of NSL gag orders that it created. Setting aside the question of whether the Termination Procedures apply to all NSLs issued before 2013, which the government has not confirmed, the fact remains that around 500,000 NSL gag orders persist. And even if the Termination Procedures applied to all of them, the FBI’s current rate of reviewing less than 600 NSL gag orders per month will not meaningfully decrease that backlog. That is why any legislative or judicial remedy for these ongoing gags must meaningfully address that hundreds of thousands of NSL recipients since 2001 still remain gagged.

Unsurprisingly, there are some limitations on our ability to extrapolate trends from the data EFF obtained. Because we do not have information about the FBI’s rate of NSL reviews under the Termination Procedures since September 2017, it is possible that the FBI has begun to close the gap since then. However, it would have to increase its pace significantly to clear the large backlog of NSLs issued before the passage of USA Freedom, not to mention whatever additional gap has developed since. As discussed further below, it should be incumbent on the government to provide further documentation of its efforts to implement the Termination Procedures, so policymakers and the public can draw more precise conclusions about their effectiveness.

C. The FBI’s data and records show that a federal appellate court was wrong to put so much weight on the Termination Procedures’ ability to remedy the unconstitutionality of the NSL statutes.

As the above shows, the FBI’s implementation of Congress’ command to review and end NSLs gag orders has proven woefully insufficient to address the First Amendment violations that result from NSL gag orders. Thus the U.S. Court of Appeals for the Ninth Circuit’s reliance on the FBI’s Termination Procedures as a means to end NSL gag orders was misplaced.

The Termination Procedures do not ensure that the duration of a gag order issued with an NSL corresponds to the particular government interests in any investigation, much less ensure that those gag orders will eventually end. Instead, as the data above shows, the FBI is only releasing 6.4 percent of NSL recipients from nondisclosure orders and instead has determined to maintain unconstitutional gag orders in the remaining 93.6 percent of the cases. Moreover, the data shows that the FBI’s low rate of reviewing NSL gag orders means that, assuming the best circumstances, it will take years before the FBI reviews the backlog of gags it has issued hundreds of thousands of times. And it remains an open question as to whether the FBI will systematically apply the Termination Procedures to NSLs issued before 2013 that do not fall within the procedure’s review criteria.

Given all of those problems, the Ninth Circuit was wrong as a legal matter to rely on the Termination Procedures to hold that the NSL statute’s content-based restriction on speech is narrowly tailored to address the government’s national security interests.43 Setting aside the question of whether the procedures can remedy the prior restraints they place on NSL recipients—which the Ninth Circuit also got wrong—the FBI data provided to EFF show that the Termination Procedures still overwhelmingly result in ongoing gag orders with no durational limits in sight. And the FBI’s implementation of the Termination Procedures shows that, left to its own discretion, the Bureau will continue to overwhelmingly uphold restrictions on NSL recipients’ speech. Thus, to the extent that the Ninth Circuit was correct in holding that the Termination Procedures themselves took on a constitutional dimension, the FBI’s practice shows that the procedures have perpetuated the unconstitutionality of the NSL statutes. The Termination Procedures do not save the statutes’ unconstitutionality. Rather, the FBI’s implementation is evidence of ongoing, widespread First Amendment violations.

The FBI and other defenders of the current NSL regime may argue that the statistics derived from the documents obtained in our FOIA suit do not support our conclusions. In particular, they may suggest that in the realm of national security it is naïve to require investigations to proceed on a “regular” schedule such that NSL nondisclosure orders can be reviewed and terminated in a timely fashion. As a result, they might take issue with the notion that the percentage of gags reviewed and terminated is “too low,” because it would be meaningless to assume there is an optimal or correct number.

In EFF’s view, these potential counterarguments only further demonstrate that the Termination Procedures represent an unworkable compromise between the FBI’s investigatory needs and the fundamental First Amendment concerns with administrative gag orders, particularly those with no fixed duration. No court has ever suggested that a fully permanent gag order would be constitutional, but the Termination Procedures explicitly allow for such permanent gags. As a policy matter, the FBI should demonstrate that the need for secrecy in an NSL gag order inheres in exactly the same way for the entirety of a years-long investigation. It defies logic to argue that a recipient cannot reveal its receipt of a years-old NSL even if the initial target is still under investigation. Moreover, by its own admission FBI uses NSLs to rule out suspects, which further suggests that the need for secrecy should not persist indefinitely by default.

IV. Why All the Silence? The Majority of NSL Recipients Don’t Publish After the FBI Gives Them Permission.

Summary

  • Companies that the FBI grants permission to speak about NSLs are largely not doing so.
  • Given the systemic failures of the NSL statutory regime and the FBI’s implementation of its duties under USA Freedom, it is all the more necessary for these private actors to take advantage of the limited speech rights granted them.

Another important lesson from EFF’s FOIA litigation is that even in the relatively few cases where the FBI officially licenses NSL recipients to speak, most continue to stay silent.44 Specifically, in contrast to tech companies like Google and Facebook, which routinely publish NSLs when given permission, Internet service providers, banks, and credit agencies as a rule do not.45

While a variety of factors presumably play into the decision to not publicize an NSL with the FBI’s permission, we should not overlook the lingering impact of a now-terminated gag order itself. From a recipient’s perspective, the NSL process begins with a demand for customer records signed by a high-ranking FBI official who legally certifies that the provider’s silence is necessary to preserving national security. Any recipient not sufficiently persuaded is reminded that failure to comply may lead to an “enforcement action,” which can include an obstruction of justice charge.46 Whether out of legal obligation, patriotism, or fear, recipients stay silent for as long as a gag order remains in place.47 Recipients have no way of knowing how long a gag will last, so they must be prepared for a period of indefinite, years-long silence.

Even when the FBI chooses to terminate a gag order, the simplest response is to do nothing. This is particularly true for companies whose business is less dependent on direct interaction with individual consumers, such as credit reporting agencies. After the Snowden revelations, many Internet companies saw an opportunity to burnish their public perception by engaging in more detailed transparency reporting.48 By contrast, banks and credit agencies may instead wager they are better served by not calling attention to the value of their business records to FBI counterintelligence and counterterrorism investigations.49 The decision not to publish is also likely self-reinforcing. Without examples of other similarly-situated companies who report on NSL gag order terminations, recipients are further conditioned to view silence as the norm.

The collective impact of NSL recipients’ failure to publicize NSLs in these cases is substantial. Above all, the public is deprived of the largest source of information about NSLs outside the government. This information could otherwise be reported, compiled, and analyzed without any harm to national security or other government interests. What’s more, the FBI and DOJ use the fact that most recipients do not speak to advocate for continued legal prohibitions on disclosure of NSLs. In opposing a constitutional challenge to the NSL statute, the DOJ argued that NSL recipients as a class are not entitled to customary First Amendment protections because they have not demonstrated a preexisting desire to speak.50 And in FOIA litigation, lawyers for FBI claimed that inaction by NSL recipients with permission to speak showed that the Bureau’s law enforcement interests might still be harmed by disclosure.51

The conclusion offered by the records EFF obtained, then, is that NSL nondisclosure orders continue to chill speech even after they terminate. Especially in light of the government’s efforts to leverage their silence for further secrecy, NSL recipients should exercise any permission they are granted to report their involvement in NSL and other national security process.52 That is not a substitute for the systemic, institutional transparency by the government we discuss in the next section, but it is the sort of corporate citizenship that should be a part of a “commitment to privacy.”53

V. Recommendations: How Congress and Courts Can End Permanent Bans On Speech.

A. Congress

When Congress passed the USA Freedom Act in 2015, it was well aware of the issue with NSL gag orders. As discussed above, the amendments in the law responded to adverse court decisions finding the NSL statute unconstitutional. According to the House Committee Report accompanying the legislation, Congress mandated the adoption of FBI’s NSL review procedures in response to a speech by President Obama, in which he argued that NSL gag orders should “terminate within a fixed time unless the government demonstrates a real need for further secrecy.”54

Along with other amendments to the NSL statute, the mandate of FBI review procedures can be seen as a compromise, which left the fundamental operation of NSLs in place, while eliminating some possibilities for the imposition of indefinite gag orders.55 However, this report demonstrates that these half-measures have been insufficient. Despite the clear intent of Congress, the FBI continues to issue new NSL gag orders with the full knowledge that they will not “terminate within a fixed time.” Congress should take several actions in response:

  1. Amend the NSL statute to prohibit the FBI from issuing gag orders without prior judicial authorization. The First Amendment prohibits administrative prior restraints, and the records EFF obtained prove that the FBI fails to cure its unconstitutional actions long after it bars NSL recipients from speaking.
  2. Ensure that the FBI initiates judicial review for all NSL gag orders on a regular and recurring basis, so that gags cannot become de facto permanent.
  3. Address existing NSL gags that are not currently subject to judicial oversight

    a. Direct FBI to review all NSLs with outstanding gag orders on a frequent basis, such as every six months.

    b. Alternatively, Congress could specify in the statute that NSL gags lasting more than two years automatically dissolve unless FBI obtains judicial order authorizing longer duration.

  4. Provide adequate funding and other resources for FBI to address the large number of outstanding NSLs, including closing underlying investigations and otherwise removing barriers to terminating nondisclosure orders.
  5. Investigate the reasons FBI has not followed the direction in the USA Freedom Act to eliminate NSL gag orders that are no longer supported by the enumerated statutory harms. This should include an independent DOJ Inspector General review provided to Congress and released to the public, in redacted form if necessary.56

A proposed revision to the NSL statute is located in the Appendix.57

B. Courts

Every court to examine the NSL statute has noted serious concerns about the authority granted to the FBI to restrict speech about the government’s law enforcement and intelligence gathering activities. Over the course of more than ten years of litigation, several federal district courts and one circuit court of appeal ruled the NSL statute unconstitutional. In response, Congress twice amended the statute, culminating with the USA Freedom Act. Even the Ninth Circuit, which upheld the post-USA Freedom version of the statute, noted that the FBI’s termination procedures “do not resolve the duration issue entirely.”58

In light of the findings described in this report, the Ninth Circuit should revisit its finding that the FBI’s procedures are sufficient to render the law narrowly tailored. In addition, district courts should follow the court’s direction to “ensure that the nondisclosure requirement does not remain in place longer than is necessary to serve the government's compelling interest.”59

Notes

  1. Doe v. Gonzales, 449 F.3d 415, 422 (2d Cir. 2006) (Cardamone, J. concurring). ↩

  2. 12 U.S.C. § 3414; 15 U.S.C. § 1681u-v; 18 U.S.C. §§ 2709, 3511; 50 U.S.C. § 3162. ↩

  3. See, e.g., 18 U.S.C. § 2709(c)(1)(B). The enumerated harms that permit nondisclosure by an NSL recipient are “(i) a danger to the national security of the United States; (ii) interference with a criminal, counterterrorism, or counterintelligence investigation; (iii) interference with diplomatic relations; or (iv) danger to the life or physical safety of any person.” ↩

  4. See In re Nat’l Sec. Letter, 930 F. Supp. 2d 1064, 1081 (N.D. Cal. 2013) (“In re NSL 2013”). ↩

  5. See 12 U.S.C. § 3414(a)(5)(A) (1986); 18 U.S.C. § 2709 (1986). ↩

  6. Among other changes, Section 505 of the PATRIOT Act significantly broadened the range of officials who could issue NSLs and lowered the standard for their issuance to allow the FBI to obtain information that is “relevant” to an “authorized investigation,” 18 U.S.C. § 2709(b) (2001).  ↩

  7. DOJ, Office of the Inspector General (“OIG”), A Review of the Federal Bureau of Investigation’s Use of National Security Letters: Assessment of Progress in Implementing Recommendations and Examination of Use in 2007 through 2009 at 65 (2014), available at https://oig.justice.gov/reports/2014/s1408.pdf (“2014 OIG Report”) (chart showing NSLs issued 2003-2011); Liberty and Security in a Changing World: Report and Recommendations from the President’s Review Group on Intelligence and Communications Technologies at 91-93 (2013) (“President’s Review Group”) (number of NSLs issued in 2012), available at http://www.whitehouse.gov/sites/default/files/docs/2013-12- 12_rg_final_report.pdf; ODNI, Statistical Transparency Report Regarding Use of National Security Authorities Calendar Year 2018 at 33 (Apr. 2019), https://www.dni.gov/files/CLPT/documents/2019_ASTR_for_CY2018.pdf (chart showing NSLs issued 2013-2018) (“2018 ODNI Transparency Report”).  ↩

  8. DOJ, OIG, A Review of the Federal Bureau of Investigation’s Use of National Security Letters 125 (2007, rereleased with some previously redacted information unredacted 2016), available at https://oig.justice.gov/reports/2016/o1601b.pdf (“2007 OIG Report”); see also 2014 OIG Report 187. ↩

  9. See, e.g., Doe v. Mukasey, 549 F.3d 861 (2d Cir. 2008), aff’g in part Doe v. Gonzales, 500 F. Supp. 2d 379 (S.D.N.Y. 2007); In re NSL 2013, 930 F. Supp. 2d at 1076; In re NSL 2017, 863 F.3d at 1119-20. ↩

  10. In re NSL 2013, 930 F. Supp. 2d at 1076. ↩

  11. See Wendy Knox Everette, The FBI Has Not Been Here (Watch Very Closely for the Removal of this Sign) Warrant Canaries and First Amendment Protection for Compelled Speech, 23 George Mason L. Rev. 377 (2016).  ↩

  12. See, e.g., Twitter v. Sessions, 263 F. Supp. 3d 803, 808 (N.D. Cal. 2017) (Twitter seeking to publish transparency report including number of NSLs received); In re Nat’l Sec. Letter, 165 F. Supp. 3d 352 (D. Md. 2015) (provider contending NSL gag no longer needed).  ↩

  13. For a detailed explanation of the changes made to the statute in response to litigation, see Charles Doyle, National Security Letters in Foreign Intelligence Investigations: A Glimpse at the Legal Background, Cong. Research Serv. (July 31, 2015), available at https://fas.org/sgp/crs/intel/RS22406.pdf. ↩

  14. Pub. L. 114-23, 129 Stat. 268 (2015) (USA FREEDOM). ↩

  15. See H.R. Rep. No. 114-109, at 26 (2015) (stating that changes to NSL nondisclosure order procedures in the Act were intended address constitutional deficiencies in the NSL statutes noted by Doe v. Mukasey, 549 F.3d 861 (2d Cir. 2008)).  ↩

  16. See Under Seal v. Sessions, Nos. 16-16067, 16-16081, 16-16082 (9th Cir.), petition for rehearing en banc pending. EFF is counsel to the NSL recipients pursuing these challenges. ↩

  17. USA FREEDOM, § 502(f). ↩

  18. Termination Procedure for National Security Letter Nondisclosure Requirement at 1, (citing 18 U.S.C. § 2709(c)) (“Termination Procedures”). ↩

  19. Id. at 2. ↩

  20. Id. ↩

  21. 18 U.S.C. § 2709(c)(1)(B). ↩

  22. Termination Procedures at 3. ↩

  23. NSL Termination Procedures at 4. The NSL termination letters produced in response to EFF’s FOIA request are available here: https://www.documentcloud.org/search/Project:%20%2217-cv-03263%20re-processed%20NSL%20termination%20letters%22. ↩

  24. See Chris Madsen, Yahoo Announces Public Disclosure of National Security Letters (June 1, 2016), https://yahoopolicy.tumblr.com/post/145258843473/yahoo-announces-public-disclosure-of-national; Richard Salgado, Sharing National Security Letters with the public (Dec. 13, 2016), https://www.blog.google/outreach-initiatives/public-policy/sharing-national-security-letters-public/; Kenneth R. Carter, Cloudflare’s Transparency Report for Second Half 2016 and an Additional Disclosure for 2013 (Jan. 10, 2017), https://blog.cloudflare.com/cloudflares-transparency-report-for-second-half-2016-and-an-additional-disclosure-for-2013-2/; Chris Sonderby, Reinforcing Our Commitment to Transparency (Sept. 30, 2016), https://newsroom.fb.com/news/2017/12/reinforcing-our-commitment-to-transparency/; Elizabeth Banker, #Transparency update: Twitter discloses national security letters (Jan. 27, 2017), https://blog.twitter.com/en_us/topics/company/2017/transparency-update-twitter-discloses-national-security-letters.html; Angela Sherrer, #Transparency update: Twitter shares more national security letters (June 20, 2018), https://blog.twitter.com/official/en_us/topics/company/2018/transparencyupdatejune2018.html.  ↩

  25. In re NSL 2017, 863 F.3d at 1126; see also Gov’t Br. at 46-47, Under Seal v. Lynch, ECF No. 52, No. 16-16082 (9th Cir. Dec. 9, 2016). ↩

  26. Id. ↩

  27. Notice Concerning National Security Letter at Issue in No 16-16067, Unsealing of Briefs and Public Identification of the Appellants, In re NSL, No. 16-16067 (9th Cir. Mar. 20, 2017), ECF No. 77. ↩

  28. See Oral Argument at 27:41, In re NSL, Nos. 16-16067 & 16-18082 (9th Cir. Mar. 22, 2017), https://youtu.be/ccS06CFkZ5M (counsel for government states that an indefinite gag order is “possible”). ↩

  29. In re NSLs, No. 16-518 (JEB), 2016 WL 7017215, at *2 (D.D.C. July 25, 2016). ↩

  30. Id. ↩

  31. Id. ↩

  32. See Barr v. Under Seal, No. 18-56669 (9th Cir. 2019). Andrew Crocker, EFF Supports Unnamed Company in Bringing an End to Endless NSL Gag Orders, EFF (May 10, 2019), https://www.eff.org/deeplinks/2019/05/eff-supports-unnamed-company-bringing-end-endless-nsl-gag-orders.  ↩

  33. EFF v. DOJ, 2019 WL 2098084, *1 (N.D. Cal. May 14, 2019). ↩

  34. Id.  ↩

  35. Id. at *2.  ↩

  36. Id.  ↩

  37. Available at https://www.eff.org/document/eff-v-doj-sept-17-2019-letter-re-nsl-termination-procedure-review-totals or Appendix at A-2 (Letter from Department of Justice to EFF, Sept. 17, 2019) (Aggregate NSL data letter). ↩

  38. Id. ↩

  39. Id. ↩

  40. In re NSLs, No. 16-518 (JEB), 2016 WL 7017215, at *2 (D.D.C. July 25, 2016). ↩

  41. As of late 2019, the Ninth Circuit was considering whether the First Amendment requires a district court to order future reviews of an NSL gag order if the court initially determines the gag is justified. Under Seal v. Barr, No. 18-56669 (9th Cir.), oral argument held Nov. 5, 2019. ↩

  42. See 2018 ODNI Transparency Report, supra note 7. In 2017, the FBI issued approximately 10,000 NSLs. ↩

  43. In re NSL 2017, 863 F.3d at 1126. ↩

  44. Cite to above analysis. ↩

  45. The documents disclosed to EFF detailing the names of roughly 750 recipients of termination letters are available at https://www.documentcloud.org/search/Project:%20%2217-cv-03263%20re-processed%20NSL%20termination%20letters%22.. EFF analyzed those documents to determine how frequently various recipients received termination letters and created a chart reflecting those totals, which is available at https://www.eff.org/document/eff-v-doj-nsl-termination-letters-recipient-totals or at Appendix A-3—A-5. ↩

  46. See 18 U.S.C. § 1510(e) (knowing and intentional violation of NSL gag order punishable by fines and up to five years in prison). ↩

  47. We are unaware of a single case in which an NSL recipient has violated an active gag order. ↩

  48. Find representative press on this. See also https://www.accessnow.org/transparency-reporting-index. ↩

  49. See Jennifer Valentino-Devries, Secret F.B.I. Subpoenas Scoop Up Personal Data From Scores of Companies, N.Y. Times (Sept. 20, 2019), https://www.nytimes.com/2019/09/20/us/data-privacy-fbi.html. (“Other companies have generally remained mum. In response to inquiries, a TransUnion spokesman would say only that the company ‘has not disclosed the receipt of any national security letters.’ A spokesman for Equifax said it was ‘compliant with the national security letters process.’”) ↩

  50. Gov’t Opp’n Brief at 58, In re NSLs, Nos. 16-16067, 16-16081, 16-16082 (9th Cir. Nov. 28, 2016). ↩

  51. See Gov’t Reply at 8, 12-13, EFF v. DOJ, ECF No. 34, No. 3:17-cv-03263-VC (N.D. Cal. Nov. 18, 2018). ↩

  52. EFF also encourages all NSL recipients to invoke the statutory “reciprocal notice” process and require the FBI to enforce NSL gags through a judicial process. See Nate Cardozo, Requiring Judicial Review for Every Gag Order Is a Simple Way to Have Our Backs, EFF (July 10, 2017), https://www.eff.org/deeplinks/2017/07/requiring-judicial-review-every-gag-order-simple-way-have-our-backs-apple-does. ↩

  53. See, e.g., Privacy, Equifax (“We have built our reputation on our commitment to. . . protect the privacy and confidentiality of personal information about consumers.”), https://www.equifax.com/privacy. ↩

  54. H. Rep. 114-109(I) (2015), 2015 U.S.C.C.A.N. 5, 26 (quoting Remarks by the President on Review of Signals Intelligence (Jan. 17, 2014), available at https://www.whitehouse.gov/the-press-office/2014/01/17/remarks-president-review-signals-intelligence.) ↩

  55. Cf. Br. of Amici Curiae Five Members of Congress, In re NSLs, Nos. 16-16067, 16-16081, 16-16082 (9th Cir. Sep. 26, 2016) (original author and sponsors of USA Freedom act arguing FBI’s review procedures as implemented do not satisfy the statute). ↩

  56. Other than a single example produced by the FBI, EFF was unable to obtain through FOIA the memoranda that FBI uses to document its decision to leave NSL gag orders in place pursuant to its review procedures. These would provide insight into whether the FBI has fulfilled its mandate to review outstanding gags and terminate those no longer supported. In an exemplar released by the FBI, the analysis and conclusion supporting continuing the gag order appears cursory. See https://www.eff.org/document/eff-v-doj-nsl-termination-letters-exemplar-memo-retain-nondisclosure-order or Appendix at A-6—A-7. ↩

  57. Available at https://www.eff.org/document/nsl-termination-procedures-proposed-rewrite-nsl-statutes or Appendix at A-8—A-10. ↩

  58. 863 F.3d at 1126. ↩

  59. Id. ↩

Downloads

PDF icon The Failed Fix to NSL Gag Orders: How the Majority of National Security Letter Recipients Remain Gagged After USA FREEDOM