Petition for Writ of Certiorari — Wikimedia Foundation, Petitioner v. National Security Agency/Central Security Service, et al.

Supreme Court briefAug 26, 2022

Ask Donna

What actually matters in this document.

Text

No. ________

IN THE

Supreme Court of the United States

WIKIMEDIA FOUNDATION ,

Petitioner,

—v.—

NATIONAL SECURITY AGENCY / CENTRAL SECURITY SERVICE, et al.,

Respondents.

ON PETITION FOR A WRIT OF CERTIORARI TO THE UNITED STATES

COURT OF APPEALS FOR THE FOURTH CIRCUIT

PETITION FOR A WRIT OF CERTIORARI

Alex Abdo

Jameel Jaffer

KNIGHT FIRST AMENDMENT

INSTITUTE AT COLUMBIA

UNIVERSITY

475 Riverside Drive, Suite 302

New York, NY 10115

Benjamin H. Kleine

Aarti Reddy

Maximilian Sladek de la Cal

COOLEY LLP

3 Embarcadero Center, 20th Fl.

San Francisco, CA 94111

Deborah A. Jeon

David R. Rocah

AMERICAN CIVIL LIBERTIES

UNION FOUNDATION

OF MARYLAND

3600 Clipper Mill Rd., #350

Baltimore, MD 21211

Patrick Toomey

Counsel of Record

Ashley Gorski

Sarah Taitz

AMERICAN CIVIL LIBERTIES

UNION FOUNDATION

125 Broad Street, 18th Floor

New York, NY 10004

(212) 549-2500

ptoomey@aclu.org

David D. Cole

AMERICAN CIVIL LIBERTIES

UNION FOUNDATION

915 15th Street, NW

Washington, DC 20005

QUESTIONS PRESENTED

1. Does the state secrets privilege described by

this Court in United States v. Reynolds and General

Dynamics Corp. v. United States authorize courts to

dismiss actions in their entirety where plaintiffs can

prove their case without reliance on privileged

evidence?

2. If the state secrets privilege does in fact

authorize courts to dismiss actions in their entirety

where plaintiffs can prove their case without

privileged evidence, may a court do so without first

determining ex parte and in camera whether the

privileged evidence establishes a valid defense?

PARTIES TO THE PROCEEDINGS

Petitioner is the Wikimedia Foundation and was

the plaintiff–appellant below.

Respondents are National Security Agency /

Central Security Service; General Paul M. Nakasone,

in his official capacity as Director of the National

Security Agency and Chief of the Central Security

Service; Office of the Director of National Intelligence;

Avril Haines, in her official capacity as Director of

National Intelligence; Department of Justice; and

Merrick B. Garland, in his official capacity as

Attorney General. They were the defendants–

appellees below.

In addition, the National Association of Criminal

Defense Lawyers, Human Rights Watch, Amnesty

International USA, PEN American Center, Global

Fund for Women, The Nation Magazine, The

Rutherford Institute, and Washington Office on Latin

America were plaintiffs in the proceedings below.

CORPORATE DISCLOSURE STATEMENT

The Wikimedia Foundation has no parent

corporation and no publicly held company owns 10

percent or more of its stock.

RELATED PROCEEDINGS

United States District Court (D. Md.):

Wikimedia Found. v. NSA, No. 1:15-cv-662 (Dec.

16, 2019)

United States Court of Appeals (4th Cir.):

Wikimedia Found. v. NSA, No. 20-1191 (Sept. 15,

2021)

iii

TABLE OF CONTENTS

QUESTIONS PRESENTED ........................................ i

PARTIES TO THE PROCEEDINGS ......................... ii

CORPORATE DISCLOSURE STATEMENT ............ ii

RELATED PROCEEDINGS....................................... ii

TABLE OF CONTENTS ............................................ iv

TABLE OF AUTHORITIES .................................... viii

PETITION FOR A WRIT OF CERTIORARI ............. 1

OPINIONS BELOW ................................................... 1

JURISDICTION.......................................................... 1

CONSTITUTIONAL AND STATUTORY

PROVISIONS INVOLVED .............................. 2

INTRODUCTION ....................................................... 2

STATEMENT OF THE CASE .................................... 6

Background ..................................................... 6

A. The Foreign Intelligence

Surveillance Act of 1978 ............................ 6

B. Warrantless Surveillance Under

Section 702 ................................................. 7

II. Statement of Facts .......................................... 8

A. Wikimedia and Its Global

Communications ........................................ 8

B. The Government’s Implementation

of Section 702 ............................................. 8

C. Upstream Surveillance .............................. 9

D. Surveillance of Wikimedia’s

Communications ...................................... 11

E. Proceedings Below ................................... 12

REASONS FOR GRANTING THE PETITION ....... 15

Whether the government may obtain

dismissal of constitutional claims under

the state secrets privilege, and what

standards control such dismissals, are

questions of extraordinary public

significance .................................................... 15

II. The circuit court cases permitting state

secrets dismissals under Reynolds

conflict with this Court’s decisions ............... 20

A. This Court’s state secrets cases do

not permit dismissal under

Reynolds where the plaintiff can

establish its case with nonprivileged

evidence .................................................... 20

B. The pre-Reynolds common law did

not permit dismissal ................................ 22

C. The circuit court precedents

allowing state secrets dismissals

directly conflict with Reynolds and

General Dynamics .................................... 24

III. The federal courts of appeals are

divided over what judicial standards

control when the government seeks

dismissal under the state secrets

privilege ......................................................... 25

A. The circuits are divided over

whether the government may obtain

a state secrets dismissal without

establishing that the privileged

evidence supports a “legally

meritorious” defense ................................ 26

B. There is broad confusion in the

lower courts about the other

circumstances in which dismissal is

available under Reynolds ........................ 30

IV. This case is an ideal vehicle to decide

whether state secrets dismissals are

permitted under Reynolds and, if they

are, what standards control .......................... 33

A. The case squarely presents the

dismissal question ................................... 33

B. The Fourth Circuit’s decision was

wrong ........................................................ 35

CONCLUSION.......................................................... 38

APPENDIX

Appendix A, Court of appeals opinion

affirming grant of summary judgment,

Sept. 15, 2021 ................................................. 1a

Appendix B, District court memorandum

opinion granting summary judgment,

Dec. 16, 2019 ................................................ 73a

Appendix C, District court order granting

summary judgment, Dec. 16, 2019 ............ 143a

Appendix D, District court memorandum

opinion denying motion to compel,

Aug. 20, 2018 .............................................. 145a

Appendix E, District court order denying

motion to compel, Aug. 20, 2018 ................ 180a

Appendix F, Court of appeals opinion

affirming in part and vacating in part

grant of motion to dismiss,

May 23, 2017 .............................................. 182a

Appendix G, District court memorandum

opinion granting motion to dismiss,

Oct. 23, 2015 ............................................... 234a

Appendix H, District court order granting

motion to dismiss, Oct. 23, 2015 ............... 272a

Appendix I, Court of appeals order denying

petition for rehearing en banc,

Mar. 29, 2022.............................................. 274a

Appendix J, 50 U.S.C. § 1881a ............................. 276a

vii

TABLE OF AUTHORITIES

Cases

Abilt v. CIA,

848 F.3d 305 (4th Cir. 2017) ................. 24, 27, 321

ACLU v. Brown,

619 F.2d 1170 (7th Cir. 1980) ............................. 33

ACLU v. NSA,

493 F.3d 644 (6th Cir. 2007) ............................... 34

Al-Haramain Islamic Found., Inc. v. Bush,

507 F.3d 1190 (9th Cir. 2007) ............................. 18

Arar v. Ashcroft,

414 F. Supp. 2d 250 (E.D.N.Y. 2006) .................. 18

Bank Line Ltd. v. United States,

68 F. Supp. 587 (S.D.N.Y. 1946) ......................... 23

Bareford v. Gen. Dynamics Corp.,

973 F.2d 1138 (5th Cir. 1992) ............................. 28

Black v. United States,

62 F.3d 1115 (8th Cir. 1995) ............................... 33

Cooke v. Maxwell,

(1817) 2 Stark. 183 .............................................. 24

Cresmer v. United States,

9 F.R.D. 203 (E.D.N.Y. 1949) .............................. 23

Duncan v. Cammell, Laird & Co.,

(1942) A.C. 624 .................................................... 24

Edmonds v. U.S. Dep’t of Just.,

323 F. Supp. 2d 65 (D.D.C. 2004) ....................... 18

viii

El-Masri v. United States,

479 F.3d 296 (4th Cir. 2007) ............. 18, 19, 27, 31

Ellsberg v. Mitchell,

709 F.2d 51 (D.C. Cir. 1983) ............................... 32

Farnsworth Cannon, Inc. v. Grimes,

635 F.2d 268 (4th Cir. 1980) ............................... 32

Fazaga v. FBI,

965 F.3d 1015 (9th Cir. 2020) ....................... 18, 28

Firth Sterling Steel Co. v. Bethlehem Steel Co.,

199 F. 353 (E.D. Pa. 1912) .................................. 22

General Dynamics Corp. v. United States,

563 U.S. 478 (2011) ...................................... passim

In re Grove,

180 F. 62 (3d Cir. 1910) ....................................... 23

Halpern v. United States,

258 F.2d 36 (2d Cir. 1958) ................................... 32

H.M.S. Bellerophon,

(1875) 44 LJR ...................................................... 23

Jewel v. NSA,

No. 08-CV-04373-JSW, 2019 WL 11504877

(N.D. Cal. Apr. 25, 2019) ..................................... 18

Kasza v. Browner,

133 F.3d 1159 (9th Cir. 1998) ........... 24, 28, 31, 33

Mohamed v. Jeppesen Dataplan, Inc.,

614 F.3d 1070 (9th Cir. 2010) ...................... passim

Molerio v. FBI,

749 F.2d 815 (D.C. Cir. 1984) ........... 24, 28, 29, 37

Monarch Assurance P.L.C. v. United States,

244 F.3d 1356 (Fed. Cir. 2001) ............................ 32

Pollen v. Ford Instrument Co.,

26 F. Supp. 583 (E.D.N.Y. 1939) ......................... 23

Ray v. Turner,

587 F.2d 1187 (D.C. Cir. 1978) ........................... 19

Redacted, 2011 WL 10945618

(FISC Oct. 3, 2011) .............................................. 12

Rex v. Watson,

(1817) 2 Stark. 116 .............................................. 23

In re Sealed Case,

494 F.3d 139 (D.C. Cir. 2007) ...................... passim

Sterling v. Tenet,

416 F.3d 338 (4th Cir. 2005) ......................... 27, 33

Tenenbaum v. Simonini,

372 F.3d 776 (6th Cir. 2004) ................... 24, 28, 32

Tenet v. Doe,

544 U.S. 1 (2005) ........................................... 17, 21

Totten v. United States,

92 U.S. 105 (1875) ................................. 4, 6, 17, 22

In re United States,

872 F.2d 472 (D.C. Cir. 1989) ....................... 29, 32

United States v. Abu Zubaydah,

142 S. Ct. 959 (2022) ........................................... 21

United States v. Pappas,

94 F.3d 795 (2d Cir. 1996) ................................... 20

United States v. Reynolds,

345 U.S. 1 (1953) .......................................... passim

Worthington v. Scribner,

109 Mass. 487 (1872) ........................................... 23

Wyatt v. Gore,

(1816) Holt N.P.C. 299 ........................................ 24

Zuckerbraun v. Gen. Dynamics Corp.,

935 F.2d 544 (2d Cir. 1991) ................. 4, 21, 28, 31

Statutes

5 U.S.C. § 552(a)(4)(B) .............................................. 19

5 U.S.C. § 552(b)(1) ................................................... 19

18 U.S.C. App. 3, Classified Information

Procedures Act (“CIPA”) ................................ 20, 25

28 U.S.C. § 1254(1) ..................................................... 1

50 U.S.C. § 1801 et seq., Foreign Intelligence

Surveillance Act (“FISA”) ............................. passim

50 U.S.C. § 1801(e) ................................................ 7

50 U.S.C. § 1801(i) ................................................. 7

50 U.S.C. § 1803(a) ................................................ 6

50 U.S.C. § 1803(i) ............................................... 25

50 U.S.C. § 1804(a) ................................................ 7

50 U.S.C. § 1805 .............................................. 7, 19

50 U.S.C. § 1805(a)(2)............................................ 7

50 U.S.C. § 1881a ...................................... 2, 7, 8, 9

50 U.S.C. § 1881a(a) .............................................. 7

50 U.S.C. § 1881a(j) ............................................... 7

Constitutional Provisions

U.S. Constitution, Fourth Amendment ................. 2, 8

Other Authorities

Final Report of the S. Select Comm. to Study

Governmental Operations with Respect to

Intelligence Activities (Book II), S. Rep.

No. 94-755 (1976)................................................... 6

Amanda Frost, The State Secrets Privilege and

Separation of Powers, 75 Fordham L. Rev.

1931 (2007) .......................................................... 18

Margaret B. Kwoka, The Procedural

Exceptionalism of National Security

Secrecy, 97 B.U. L. Rev. 103 (2017) .................... 18

Office of Dir. of Nat’l Intel., Annual Statistical

Transparency Report 17 (Apr. 2022) .................... 9

William Sanford, Evidentiary Privileges

Against the Production of Data Within

the Control of Executive Departments,

3 Vanderbilt L. Rev. 73 (1949) ............................ 23

John Henry Wigmore, Evidence in Trials at

Common Law § 2212a (3d ed. 1940) ................... 16

xii

PETITION FOR A WRIT OF CERTIORARI

Petitioner respectfully requests a writ of certiorari

to review the judgment of the United States Court of

Appeals for the Fourth Circuit.

OPINIONS BELOW

The opinion of the court of appeals (App. 1a) is

reported at Wikimedia Foundation v. NSA, 14 F.4th

276 (4th Cir. 2021). The opinion of the district court

granting summary judgment (App. 73a) is reported at

427 F. Supp. 3d 582 (D. Md. 2019). The opinion of the

district court denying Petitioner’s motion to compel

(App. 145a) is reported at 335 F. Supp. 3d 772 (D. Md.

2018). The opinion of the court of appeals vacating, in

part, the district court’s dismissal of the complaint

(App. 182a) is reported at 857 F.3d 193 (4th Cir. 2017).

The opinion of the district court granting

Respondents’ motion to dismiss (App. 234a) is

reported at 143 F. Supp. 3d 344 (D. Md. 2015).

JURISDICTION

The court of appeals entered judgment on

September 15, 2021 (App. 1a), and it denied rehearing

on March 29, 2022 (App. 274–75a). On June 6, 2022,

this Court extended the time within which to file a

petition for a writ of certiorari to August 26, 2022.

This Court has jurisdiction under 28 U.S.C.

§ 1254(1).

CONSTITUTIONAL AND STATUTORY

PROVISIONS INVOLVED

The Fourth Amendment to the U.S. Constitution

provides:

The right of the people to be secure in

their persons, houses, papers, and

effects, against unreasonable searches

and seizures, shall not be violated, and

no Warrants shall issue, but upon

probable cause, supported by Oath or

affirmation, and particularly describing

the place to be searched, and the persons

or things to be seized.

The relevant statutory provision, 50 U.S.C.

§ 1881a, is set forth at App. 276–305a.

INTRODUCTION

This lawsuit challenges the suspicionless search

and seizure of U.S. persons’ international Internet

communications by the National Security Agency

(“NSA”). As the government has publicly

acknowledged, the NSA systematically searches

communications flowing into and out of the United

States on the Internet’s central arteries, looking for

information relating to thousands of foreignintelligence surveillance targets. This “Upstream”

surveillance involves a sweeping invasion of

Americans’ privacy. It is the digital analogue of the

government reading every letter that comes through

mail processing centers in order to determine which

letters to keep.

Although this mass surveillance of Americans’

private communications raises grave constitutional

questions, its lawfulness has yet to be considered by

any ordinary court, civil or criminal, in the more than

twenty years of its operation.

Petitioner filed this lawsuit in 2015 challenging

the legality of Upstream surveillance. A divided

Fourth Circuit panel held that Petitioner had

presented sufficient evidence of its standing, based on

public disclosures, to make out a prima facie case and

defeat summary judgment. But citing the state secrets

privilege recognized by this Court in United States v.

Reynolds, 345 U.S. 1 (1953), the panel held that the

mere possibility the government could have secret

evidence that might provide a defense required

dismissal. App. 55–58a. The court did not conduct an

ex parte, in camera review to determine whether that

evidence in fact existed, let alone whether it would

establish a valid defense.

Judge Motz dissented. She observed that the

majority’s approach to the state secrets privilege and

its reliance on “far-fetched hypotheticals” represented

a “dramatic departure” from Reynolds, which warned

that “[j]udicial control over the evidence in a case

cannot be abdicated to the caprice of executive

officers.” App. 62a (quoting 345 U.S. at 9–10). And she

noted that the court’s ruling conflicted with other

circuits. App. 62a.

Whether the executive branch can invoke the state

secrets privilege to dismiss cases in their entirety

where the plaintiff can make its case without

privileged evidence—and, if so, under what

circumstances—are questions of extraordinary

importance. Since Reynolds was decided more than 70

years ago, the government’s increasingly broad and

categorical assertions of privilege have prevented

courts from reviewing the constitutionality of

executive branch conduct across a wide range of

contexts.

This case presents two significant conflicts that

only this Court can resolve.

First, the circuit court cases permitting state

secrets dismissals where plaintiffs can make their

case using nonprivileged information conflict with

this Court’s decisions in Reynolds and General

Dynamics Corp. v. United States, 563 U.S. 478 (2011).

As these decisions emphasize, the state secrets

privilege is an evidentiary privilege. When

successfully invoked, it results in the exclusion of the

privileged evidence, leaving the parties to proceed

with unprivileged evidence. The lower courts,

however, have improperly treated the Reynolds

privilege as a license to dismiss cases altogether—

even where plaintiffs can establish liability based on

nonprivileged evidence. This approach wrongly

conflates the Reynolds evidentiary privilege with the

narrow justiciability bar that the Court recognized in

Totten v. United States, 92 U.S. 105 (1875). But the

Totten state secrets doctrine is “quite different,” Gen.

Dynamics, 563 U.S. at 48, and rests on a distinct legal

foundation based on the assumption of risk when one

enters into a secret government contract.

Second, even if state secrets dismissals are

permitted under Reynolds where a plaintiff can make

out its case with nonprivileged evidence, the circuit

courts are in conflict over the standards that govern

such dismissals. The D.C. and Ninth Circuits have

held that, to obtain dismissal on the ground that the

government’s defense would implicate state secrets,

the government must establish through an ex parte,

in camera procedure that the defense is “valid”—

meaning the privileged evidence exists and supports a

legally meritorious defense. The Sixth and Second

Circuits have endorsed this approach as well.

The Fourth Circuit, by contrast, permits the

government to obtain dismissal based on the

government’s assertion that “hypothetical” defenses

would implicate state secrets—without requiring the

government to submit evidence establishing any

defense, even in an ex parte and in camera

submission. Under this approach, even if the plaintiff

can make out its case with nonprivileged evidence,

and the defendant has no valid defense—meaning

that the challenged conduct is indeed illegal—the

court will nonetheless dismiss the suit on privilege

grounds. As applied here, that means that even

assuming that Upstream surveillance is an

indefensible violation of Petitioner’s constitutional

rights, the case will be dismissed and the

constitutional violations will continue.

The Court should grant review to decide whether

state secrets dismissals are permitted where a

plaintiff can prove its case using nonprivileged

evidence, and if so, whether the courts must

determine that the defense asserted is meritorious, in

both fact and law.

STATEMENT OF THE CASE

Background

The Foreign Intelligence

Surveillance Act of 1978

In 1975, Congress established a committee,

chaired by Senator Frank Church, to investigate

allegations of “substantial wrongdoing” by the

intelligence agencies in their conduct of surveillance.

Final Report of the S. Select Comm. to Study

Governmental

Operations

with

Respect

Intelligence Activities (Book II), S. Rep. No. 94-755, at

v (1976). The Committee discovered that the

intelligence agencies had, for decades, “infringed the

constitutional rights of American citizens” and

“intentionally

disregarded”

limitations

surveillance in the name of “national security.” Id. at

137. The agencies had “pursued a ‘vacuum cleaner’

approach to intelligence collection,” intercepting

Americans’ communications under the pretext of

targeting foreigners. Id. at 165. The Committee

recommended that all surveillance of communications

“to, from, or about an American without his consent”

be subject to a warrant procedure. Id. at 309.

In 1978, Congress responded by enacting the

Foreign Intelligence Surveillance Act (“FISA”), which

regulates surveillance conducted for foreign

intelligence purposes. The statute created the Foreign

Intelligence Surveillance Court (“FISC”) and

empowered it to review applications for surveillance

in certain foreign intelligence investigations. 50

U.S.C. § 1803(a). As originally enacted, FISA

generally required the government to obtain an

individualized order from the FISC based on a

detailed factual showing before conducting electronic

surveillance on U.S. soil. Id. §§ 1804(a), 1805. The

FISC could authorize surveillance only if it found that,

among other things, there was “probable cause to

believe that the target of the electronic surveillance

[was] a foreign power or an agent of a foreign power.”

Id. § 1805(a)(2).

Warrantless Surveillance Under

Section 702

Section 702 of FISA, enacted in 2008, radically

altered the FISA regime in two key ways.

First, Section 702 allows the surveillance of U.S.

persons’ international communications without a

warrant or any individualized court approval. 1

Instead, the FISC merely reviews, on an annual basis,

the general procedures used in conducting the

surveillance. 50 U.S.C. § 1881a(j).

Second, Section 702 authorizes surveillance not

predicated on any suspicion of wrongdoing. The

statute permits the government to target any

foreigner located outside the United States to obtain

“foreign intelligence information,” which is defined

broadly to include any information bearing on the

foreign affairs of the United States. Id. §§ 1881a(a),

1801(e). The government’s targets need not be agents

of foreign powers, terrorists, or suspected of criminal

activity.

Consistent with FISA, Petitioner uses the phrase “U.S.

persons” to refer to U.S. citizens and residents. 50 U.S.C.

§ 1801(i). Petitioner uses the term “international” to describe

communications that either originate or terminate outside the

United States, but not both.

Section 702 exposes every communication between

an individual in the United States and a nonAmerican abroad to potential surveillance. And the

government is using the statute to conduct precisely

the kind of vacuum-cleaner-style surveillance that the

Church Committee condemned and that the Fourth

Amendment was intended to prohibit.

II.

Statement of Facts

Wikimedia and Its Global

Communications

As the operator of one of the most-visited websites

in the world, Wikimedia, a U.S. non-profit

organization, engages in more than one trillion

international Internet communications each year.

JA.3: 2255, 2264. 2 Wikimedia communicates with

hundreds of millions of people in every country on

Earth—as they read, edit, and contribute to the

twelve Wikimedia “Projects.” JA.3: 2264, 2220–31

best-known project is Wikipedia, a free Internet

encyclopedia that is one of the largest collections of

shared knowledge in human history. Wikimedia’s

communications are essential to its organizational

mission, as is its ability to protect the privacy of these

communications. JA.3: 2235, 2242.

The Government’s Implementation

of Section 702

The government has relied on Section 702 to

intercept and retain huge volumes of Americans’

communications. Privacy & Civil Liberties Oversight

2 “JA” citations are to the joint appendix and volume number

as filed in the court of appeals, No. 20-1191, on July 1, 2020.

Board (“PCLOB”) Report 152 (JA.4: 2591). Each year,

the NSA targets more than 100,000 individuals and

groups for surveillance under Section 702. JA.4: 2762;

see also Office of Dir. of Nat’l Intel., Annual Statistical

Transparency Report 17 (Apr. 2022) (reporting

232,432 targets in 2021), https://perma.cc/GL5C5DJH. Whenever a U.S. person communicates with

any one of the government’s targets, his or her

communications can be intercepted and retained. In

2011 alone, the government relied on Section 702 to

intercept and retain more than 250 million

communications—a number that does not reflect the

far larger quantity of communications whose contents

the NSA searched and then discarded. PCLOB Report

37, 111 n.476, 116 (JA.4: 2476, 2550, 2555).

Upstream Surveillance

This case concerns Upstream surveillance, which

involves the government’s warrantless search and

seizure of U.S. persons’ Internet communications on

U.S. soil pursuant to Section 702. PCLOB Report 36–

41 (JA.4: 2475–80). The government has disclosed a

significant amount of information about Upstream

surveillance, including dozens of FISC opinions and

filings, an exhaustive report by the Privacy and Civil

Liberties Oversight Board, public testimony by

intelligence officials, and official statements by the

NSA. See, e.g., JA.4: 2436; Off. of the Director of Nat’l

Intel., Document Release (JA.4: 2718).

To conduct Upstream surveillance, the NSA

intercepts communications that transit Internet

“backbone” circuits—the “high-speed, ultra-high

bandwidth” Internet circuits operated by major

communication service providers. PCLOB Report 36–

37 (JA.4: 2475–76); JA.4: 2739; 2d Bradner Decl.

¶¶ 13–16 (JA.7: 3887–88). The NSA scans

international Internet communications to find

“selectors”—such as email addresses or phone

numbers—associated with its many targets. PCLOB

Report 37–41 (JA.4: 2476–80); JA.4: 2729–30, 2737–38.

The breadth of Upstream surveillance is a

function, in large part, of how communications

traverse the Internet. When an individual engages in

any kind of Internet activity, such as browsing a

webpage or sending an email, her communications are

broken up into data “packets”—small chunks of

information. Bradner Decl. ¶ 49 (JA.2: 941). These

packets are transmitted separately across the

Internet circuits described above, and during their

journey, they are mixed up with the packets of

countless other communications. Id. ¶ 104 (JA.2: 959).

As a result, the NSA cannot know in advance

which packets belong to communications to or from its

targets. 2d Bradner Decl. ¶¶ 54–58, 68–70, 75–84

(JA.7: 3898–3900, 3903–04, 3906–09). Instead, to

identify the communications of its targets on any

particular circuit, the NSA must copy all packets of

potential interest, reassemble those packets into

communications, and search those communications

for selectors. Id. ¶ 55 (JA.7: 3899).

The government’s disclosures make clear that

Upstream surveillance involves: (1) the copying of

packets on a circuit; (2) the reassembly of packets into

“transactions”; (3) the review of those transactions for

the presence of selectors associated with its

surveillance targets; and (4) the ingestion of

transactions that contain selectors into the NSA’s

databases. Id. ¶¶ 13–16 (JA.7: 3887–88); Bradner

Decl. ¶¶ 6(a)–(c), 250–330 (JA.2: 926, 1012–40). 3

In some instances, the NSA filters the stream of

communications to eliminate packets that are wholly

domestic, prior to reassembly and review. Bradner

Decl. ¶¶ 290–94 (JA.2: 1025–27). But significantly

here, the government does not perform any filtering

when it conducts Upstream surveillance at

“international Internet links.” 2d Bradner Decl.

¶¶ 25(e), 35, 42–45 (JA.7: 3890, 3893–95). Instead, the

NSA is copying and searching all of the

communications on the international links it is

monitoring. 2d Bradner Decl. ¶ 43 (JA.7: 3895).

Surveillance of Wikimedia’s

Communications

As Wikimedia’s expert, Scott Bradner, explains, it

is “virtually certain that the NSA has, in the course of

the upstream collection program, copied, reassembled

and reviewed at least some of Wikimedia’s

communications.” Bradner Decl. ¶ 6(e) (JA.2: 927); 2d

Bradner Decl. ¶¶ 3–4 (JA.7: 3883). That conclusion

flows from three central facts:

First, Wikimedia’s trillions of communications

traverse every international Internet link carrying

public Internet traffic into and out of the United

3 Until April 2017, the NSA ingested communications that

were to, from, or “about” a targeted selector. FISC Mem. Op. &

Order 16 (Apr. 26, 2017) (JA.4: 2806). In April 2017, the NSA

chose to suspend “about” collection after disclosing that, for

years, it had violated court-ordered rules intended to protect

Americans’ privacy. Id. at 19–23 (JA.4: 2809–13).

States. Bradner Decl. ¶¶ 6(d), 336–38, 341–50 (JA.2:

927, 1043–47).

Second, the NSA conducts Upstream surveillance

on at least one “international Internet link.” Redacted,

2011 WL 10945618, at *15 (FISC Oct. 3, 2011);

Bradner Decl. ¶¶ 225, 331–34 (JA.2: 1003, 1040–42).

Third, based on the government’s disclosures

about the operation, breadth, and goals of Upstream

surveillance, the NSA is copying and reviewing some

of Wikimedia’s communications on every link it is

monitoring. 2d Bradner Decl. ¶¶ 17–155 (JA.7: 3888–

3938); Bradner Decl. ¶¶ 6–7, 250–370 (JA.2: 926–27,

1012–60).

Proceedings Below

In 2015, Wikimedia and eight others sued

Defendants, claiming that Upstream surveillance

violates the Constitution and FISA. The district court

dismissed the suit for lack of standing. The Fourth

Circuit vacated the district court’s order as to

Wikimedia, holding that Wikimedia had plausibly

alleged that it was subject to Upstream surveillance.

App. 211–12a.

On remand, the district court ordered

jurisdictional discovery on Wikimedia’s standing. The

parties engaged in paper discovery and Wikimedia

conducted a day-long deposition of an NSA witness

that provided unclassified information about

Upstream surveillance. See, e.g., JA4: 2721; JA.1:

0287.

The government then moved for summary

judgment, arguing that Wikimedia did not have

standing and that, in any event, the state secrets

privilege required dismissal of the entire case.

In support of both arguments, the government

raised a set of hypothetical defenses. It relied

principally on the expert declarations of Henning

Schulzrinne, who opined that, as a theoretical matter,

it would be possible to design a system of Upstream“style” surveillance that deliberately ignored every

one of Wikimedia’s trillions of communications. JA.1:

759. He did not claim that the government had

actually done so nor provide any evidence supporting

his hypotheticals, and he admitted that he had “no

knowledge” of the NSA’s practices. Schulzrinne Decl.

¶ 53 (JA.2: 743); see App. 16a. Nonetheless, the

government argued that further litigation would

reveal state secrets. In response, Wikimedia relied

principally on the expert declarations of Scott

Bradner. After reviewing the government’s extensive

public disclosures in light of fundamental network

engineering principles, Bradner explained it was

“virtually certain” that Wikimedia’s communications

were subject to the NSA’s surveillance. JA.7: 3938.

Wikimedia also argued that the state secrets privilege

did not entitle the government to dismissal.

The district court granted the government’s motion

for summary judgment, App. 143a, holding that

Wikimedia had not established a genuine dispute of

material fact as to its standing, and that further

litigation was barred by the state secrets privilege. Id.

On appeal, Judges Diaz and Motz held that

Wikimedia had presented sufficient evidence, based

on the government’s official disclosures and the

Bradner declarations, on which a reasonable

factfinder could find standing. App. 25–35a.

Judges Diaz and Rushing, however, affirmed the

district court’s dismissal based on state secrets. App.

58a. The court accepted the government’s claim that,

in theory, it could have a secret defense—one that

relied on privileged evidence and would be “central” to

the case. App. 55–58a. The court did not conduct an in

camera review to determine whether such evidence

existed or whether it would in fact establish a

meritorious defense.

Judge Motz dissented from the court’s state secrets

ruling, describing the majority opinion as a “dramatic

departure” from Reynolds. App. 62a. Her dissent

expressed “serious concerns” with the majority’s

willingness to dismiss the suit based on “far-fetched

hypotheticals,” and its “relegat[ion] [of] the judiciary

to the role of a bit player in cases where weighty

constitutional interests ordinarily require us to cast a

more ‘skeptical eye.’” App. 62a (quoting Abilt v. CIA,

848 F.3d 305, 312 (4th Cir. 2017)). Judge Motz

explained that under Reynolds, “in camera review is a

‘necessary process’ when, as here, the Government

asserts that the state secrets privilege will preclude it

from raising a valid defense to a constitutional claim.”

App. 64a (quoting Molerio v. FBI, 749 F.2d 815, 825

(D.C. Cir. 1984) (Scalia, J.)).

Wikimedia’s petition for rehearing en banc was

denied on March 29, 2022. App. 274–75a.

REASONS FOR GRANTING THE PETITION

Whether the government may obtain

dismissal of constitutional claims under

the state secrets privilege, and what

standards control such dismissals, are

questions of extraordinary public

significance.

As this Court recognized in United States v.

Reynolds, 345 U.S. 1 (1953), and reaffirmed in General

Dynamics Corp. v. United States, 563 U.S. 478 (2011),

the state secrets privilege is an evidentiary privilege.

Yet, as this case illustrates, some lower courts have

transformed it into an immunity doctrine that

effectively places many government policies beyond

the reach of the Constitution, even where the plaintiff

can establish liability based entirely on nonprivileged

information. The questions presented by this case—

whether the state secrets privilege entitles the

government to the dismissal of an action in its entirety

and, if so, when—are of extraordinary importance,

with profound implications for the rule of law.

In Reynolds, the relatives of three civilians who

died in the crash of a military plane in Georgia sued

the manufacturer for damages. In response to a

discovery request for the flight accident report, the

government asserted the state secrets privilege,

arguing that the report contained information about

secret military equipment that was being tested

aboard the aircraft during the fatal flight. 345 U.S. at

3–4. Noting that the government’s privilege to resist

discovery of “military and state secrets” was “not to be

lightly invoked,” the Court required “a formal claim of

privilege, lodged by the head of the department which

has control over the matter, after actual personal

consideration by that officer.” Id. at 7–8. The greater

the necessity for the allegedly privileged information

in presenting the case, the more a “court should probe

in satisfying itself that the occasion for invoking the

privilege is appropriate.” Id. at 11. The Reynolds

Court cautioned that “judicial control over the

evidence in a case cannot be abdicated to the caprice

of executive officers.” Id. at 9–10.

The Court emphasized that the privilege was “well

established in the law of evidence,” 345 U.S. at 6–7,

and cited treatises, including Wigmore’s Evidence in

Trials at Common Law, as authority. Wigmore

acknowledged that there “must be a privilege for

secrets of State.” 8 John Henry Wigmore, Evidence in

Trials at Common Law § 2212a (3d ed. 1940). He

cautioned, however, that the privilege “has been so

often improperly invoked and so loosely misapplied

that a strict definition of its legitimate limits must be

made.” Id. That included requiring the trial judge to

scrutinize closely the evidence at issue:

Shall every subordinate in the department

have access to the secret, and not the

presiding officer of justice? Cannot the

constitutionally coördinate body of

government share the confidence? The

truth cannot be escaped that a Court

which abdicates its inherent function of

determining the facts upon which the

admissibility of evidence depends will

furnish to bureaucratic officials too ample

opportunities for abusing the privilege.

Id. at § 2379.

In General Dynamics, the Court considered the

application of the state secrets privilege to a contract

dispute between the government and contractors that

had agreed to develop a stealth aircraft for the Navy.

563 U.S. at 480. This Court held that the case was

governed not by the state secrets privilege, as the

lower courts had held, but by the “quite different” rule

announced in Totten, 92 U.S. at 105, and Tenet v. Doe,

544 U.S. 1 (2005). That rule involves “alleged

contracts to spy.” 563 U.S. at 485–86. In rejecting the

applicability of the Reynolds state secrets privilege,

the Court reiterated that that privilege is an

evidentiary one:

Reynolds was about the admission of

evidence. It decided a purely evidentiary

dispute by applying evidentiary rules:

The privileged information is excluded

and the trial goes on without it.

Id. at 485.

Both Reynolds and General Dynamics thus

emphasize that the state secrets privilege is

evidentiary in nature, and that its effect is only to

exclude the relevant evidence from the litigation—

“and the trial goes on without it.” Id. If the plaintiff

requires the privileged information to make its case,

the suit will be dismissed. But otherwise, the case

proceeds without the excluded evidence, as it would

with any other evidentiary privilege.

In recent years, however, lower courts have

fundamentally expanded the privilege beyond its

evidentiary foundation. Courts have relied on it to

permit the government to obtain dismissal of cases,

including those where the plaintiff can make its case

with exclusively nonprivileged evidence. And the

government has invoked the privilege more often, and

in cases of greater national significance. 4 When so

transformed, the privilege ceases to be merely

evidentiary, and obstructs challenges even to

unconstitutional conduct.

The government has invoked the privilege to

terminate a whistleblower suit brought by a former

FBI translator who was fired after reporting serious

security breaches and possible espionage within the

Bureau. Edmonds v. U.S. Dep’t of Just., 323 F. Supp.

2d 65 (D.D.C. 2004), cert. denied, 74 USLW 3108 (U.S.

Nov. 28, 2005) (No. 05-190). It invoked the privilege to

seek dismissal of suits challenging the government’s

seizure, transfer, and torture of foreign citizens. See

El-Masri v. United States, 479 F.3d 296 (4th Cir.

2007); Mohamed v. Jeppesen Dataplan, Inc., 614 F.3d

1070, 1076 (9th Cir. 2010) (en banc); Arar v. Ashcroft,

414 F. Supp. 2d 250 (E.D.N.Y. 2006) (dismissed on

other grounds). And it has repeatedly invoked the

privilege seeking to foreclose judicial review of the

NSA’s warrantless surveillance of United States

citizens. See, e.g., Jewel v. NSA, No. 08-CV-04373JSW, 2019 WL 11504877, at *2 (N.D. Cal. Apr. 25,

2019), aff’d, 856 F. App’x 640 (9th Cir. 2021); AlHaramain Islamic Found., Inc. v. Bush, 507 F.3d

1190, 1193 (9th Cir. 2007); see also Fazaga v. FBI, 965

Margaret B. Kwoka, The Procedural Exceptionalism of

National Security Secrecy, 97 B.U. L. Rev. 103, 118 (2017)

(describing a “steep increase in the invocation of the state secrets

privilege”); Amanda Frost, The State Secrets Privilege and

Separation of Powers, 75 Fordham L. Rev. 1931, 1939 (2007)

(“The Bush Administration has raised the privilege in twentyeight percent more cases per year than in the previous decade,

and has sought dismissal in ninety-two percent more cases per

year than in the previous decade.”).

F.3d 1015, 1024–25 (9th Cir. 2020), rev’d, 142 S. Ct.

1051 (2022) (invoking the state secrets privilege to

dismiss claims challenging FBI surveillance based on

religion).

In these cases, the government sought dismissal,

not merely the exclusion of evidence, often even before

the plaintiff had an opportunity to prove its case with

unprivileged information. As a result, a broad range

of government action has been shielded from judicial

review. In effect, the government has used the

privilege to declare these cases nonjusticiable—

without producing specific privileged evidence,

without having to justify its claims by reference to

specific facts that would be necessary and relevant to

adjudicate the case, and without having to submit its

claims to even modified adversarial testing. And it has

done so even in cases where the plaintiff was prepared

to proceed on the basis of nonprivileged evidence. See,

e.g., El-Masri, 479 F.3d at 296; Jeppesen, 614 F.3d at

1075.

The lower courts have adopted this course of

deferential dismissals on state secrets grounds

despite the fact that in the 70 years since Reynolds,

judges have become more accustomed to assessing

claims regarding sensitive and classified information,

and are better equipped to do so than when this Court

decided Reynolds. Under the Freedom of Information

Act, for instance, courts routinely determine whether

the government has properly classified information.

See 5 U.S.C. § 552(a)(4)(B), (b)(1); Ray v. Turner, 587

F.2d 1187, 1191–95 (D.C. Cir. 1978). Under FISA,

Article III judges review highly sensitive information

ex parte and in camera to determine whether

surveillance is lawful. See 50 U.S.C. §§ 1805, 1806(f).

And the Classified Information Procedures Act

(“CIPA”), 18 U.S.C. App. 3, empowers federal judges

to employ special procedures for the introduction and

handling of classified information in certain kinds of

trials. See United States v. Pappas, 94 F.3d 795, 799

(2d Cir. 1996). Congress has clearly recognized that

courts can and sometimes must review classified

materials, and courts have demonstrated that they

can securely handle secret information in the context

of litigation.

The lower courts’ radical transformation of the

Reynolds privilege from an evidentiary privilege to a

liability shield, even where serious constitutional

violations are alleged, warrants this Court’s review.

II.

The circuit court cases permitting state

secrets dismissals under Reynolds

conflict with this Court’s decisions.

Review is also warranted because the court below,

and several others, have now applied the state secrets

privilege in a manner that conflicts with Reynolds and

General Dynamics. This Court has never authorized

dismissal under Reynolds, and the pre-Reynolds

common law did not permit dismissal where a plaintiff

could establish a prima facie case with nonprivileged

material.

This Court’s state secrets cases do

not permit dismissal under

Reynolds where the plaintiff can

establish its case with

nonprivileged evidence.

This Court has established two distinct state

secrets doctrines, which the court below, and several

other courts, have erroneously conflated.

The Reynolds privilege is a common-law privilege

“in the law of evidence,” concerning “military and

state secrets.” Reynolds, 345 U.S. at 6–7. Where it

applies, it operates like other evidentiary privileges:

“[t]he privileged information is excluded and the trial

goes on without it.” Gen. Dynamics, 563 U.S. at 485.

As the Court summarized just last term, Reynolds

“allows the Government to bar the disclosure of

information that, were it revealed, would harm

national security.” United States v. Abu Zubaydah,

142 S. Ct. 959, 963 (2022).

Crucially, however, even if a court approves

invocation of the Reynolds privilege, “the Court [does]

not order judgment in favor of the Government.” Gen.

Dynamics, 563 U.S. at 485; see Reynolds, 345 U.S. at

11. Instead, this Court has directed that plaintiffs be

given the opportunity to make their case without the

privileged evidence. Id. Only if a plaintiff fails to do so

should the suit be dismissed.

The other “quite different” state secrets doctrine is

a justiciability bar. Gen. Dynamics, 563 U.S. at 485.

It is specific to disputes over secret government

contracts, and requires dismissal “where the very

subject matter of the action . . . [i]s a matter of state

secret.” Tenet, 544 U.S. at 9; id. at 12 (Scalia, J.,

concurring) (“[T]he bar of Totten is a jurisdictional

one.”). General Dynamics reaffirmed that this

jurisdictional bar is no “mere evidentiary point,” and

that the authority for it arises “not [from] Reynolds,

but [from] two cases dealing with alleged contracts to

spy.” Gen. Dynamics, 563 U.S. at 485–86 (citing Totten

and Tenet).

The Totten doctrine rests on the Court’s “authority

to fashion contractual remedies in Governmentcontracting disputes.” Gen. Dynamics, 563 U.S. at

485. In both Totten and Tenet, the Court held the

contracts to be unenforceable because the parties to

the contracts “must have understood that the lips of

the other were to be for ever sealed,” Totten, 92 U.S.

at 106, and because judicial “refusal to enforce [the

contracts] captures what the ex ante expectations of

the parties were or reasonably ought to have been,”

Gen. Dynamics, 563 U.S. at 490.

Outside the Totten line of cases, however, the state

secrets privilege does not warrant dismissal where the

plaintiff can prove its case without privileged

evidence. “The privileged information is excluded and

the trial goes on without it.” General Dynamics. 563

U.S. at 485.

The pre-Reynolds common law did

not permit dismissal.

A comprehensive review of the pre-Reynolds

common law authorities, including those cited by

Reynolds, confirms that they would not permit

dismissal in cases like this one. With the exception of

Totten, in every state secrets case that Reynolds cites,

and in every case on which those cases rely, if the

privilege was sustained, the evidence was excluded

and the case proceeded without it. Courts dismissed

cases only when the plaintiffs could not make out their

case without the excluded evidence.

The American cases Reynolds cites confirm this.

See Firth Sterling Steel Co. v. Bethlehem Steel Co., 199

F. 353, 356 (E.D. Pa. 1912) (court order “expunging

the exhibits in question from the record,” no

dismissal); Pollen v. Ford Instrument Co., 26 F. Supp.

583, 586 (E.D.N.Y. 1939) (motion to compel denied, no

dismissal); Cresmer v. United States, 9 F.R.D. 203, 204

(E.D.N.Y. 1949) (court conducted ex parte, in camera

review in response to privilege assertion and granted

motion to compel); Bank Line Ltd. v. United States, 68

F. Supp. 587, 588 (S.D.N.Y. 1946) (reaffirming prior

order granting motion to compel production of Navy

record); In re Grove, 180 F. 62, 70 (3d Cir. 1910)

(reversing contempt order where witness had refused

to produce documents deemed secret by the Navy,

without regard to underlying litigation).

The secondary sources cited in Reynolds also

confirm that the privilege is solely evidentiary.

Sanford described the privilege as applicable where

“[a] public interest demands that such matters be

beyond the reach of court processes for production or

disclosure,” without mentioning dismissal as a

remedy. William Sanford, Evidentiary Privileges

Against the Production of Data Within the Control of

Executive Departments, 3 Vanderbilt L. Rev. 73, 75

(1949). No example cited in Sanford, Wigmore, or

Greenleaf involves dismissal where the plaintiff could

prove its case without privileged evidence. See, e.g.,

H.M.S. Bellerophon, [1875] 44 LJR 5–9 (cited in

Wigmore) (excluding evidence and then resolving the

merits for defendants); Rex v. Watson, [1817] 2 Stark.

116, 148, 159 (excluding evidence defendant sought;

defendant ultimately acquitted); Worthington v.

Scribner, 109 Mass. 487 (1872) (cited in Greenleaf)

(extensively reviewing English and American

privilege cases without ever mentioning dismissal,

and concluding that reports of potential criminal

activity could not be sought by interrogatory).

And the English common law cases Reynolds cites

are in accord. Duncan v. Cammell, Laird & Co., [1942]

A.C. 624 (H.L.), upheld a privilege claim against a

subpoena for documents in a private suit, without

dismissal. Duncan cited many other cases, none of

which support dismissal here. See, e.g., Wyatt v. Gore,

[1816] Holt N.P.C. 299, 305 (plaintiff won even after

privileged material was excluded); Cooke v. Maxwell,

[1817] 2 Stark. 183, 185–86 (same).

The circuit court precedents

allowing state secrets dismissals

directly conflict with Reynolds and

General Dynamics.

Notwithstanding the clear conceptual difference

between Reynolds and Totten, the court below, and

several other circuit courts, have erroneously

conflated the two by permitting dismissal based on the

Reynolds privilege, even if a plaintiff could prove its

case using only public evidence.

For example, in the decision below, the Fourth

Circuit ordered dismissal even though Petitioner is

prepared to make its case on nonprivileged evidence,

because the court believed that the government’s

defense would implicate privileged evidence. App. 56–

57a; see also Abilt, 848 F.3d at 314 (summarizing

circuit precedent). The D.C., Sixth, and Ninth Circuits

have similarly dismissed suits where excluding

privileged evidence from the suit would deprive the

defendant of a “valid defense.” Molerio, 749 F.2d at

825 (dismissing suit after reviewing the privileged

evidence in camera and determining that the

plaintiff’s suit was meritless); Kasza v. Browner, 133

F.3d 1159, 1166 (9th Cir. 1998); Tenenbaum v.

Simonini, 372 F.3d 776, 777 (6th Cir. 2004). And,

more generally, the Ninth Circuit has countenanced

dismissal at the pleading stage of cases that do not

implicate government contracts based on its mistaken

view that “the Totten bar and the Reynolds privilege

form a continuum of analysis.” Jeppesen, 614 F.3d at

1089.

All of these decisions conflict with General

Dynamics, which made clear that the Totten bar and

the Reynolds privilege are “quite different.” 563 U.S.

at 485. Where the plaintiff can establish a prima facie

case of liability without privileged evidence, there is

no basis for ruling in defendant’s favor. If the

government needs privileged evidence to defend its

conduct, it can raise that defense using specialized

procedures—e.g., in camera proceedings with

security-cleared counsel—just as it does under CIPA,

alongside security-cleared amici in the FISC, and in

hundreds of habeas cases arising out of Guantánamo.

See 18 U.S.C. App. 3; 50 U.S.C. § 1803(i). But allowing

the government to prevent judicial consideration of

challenges to grave constitutional abuses even where

a plaintiff can make a prima facie case would allow it

to escape accountability by violating the Constitution

in secret.

III.

The federal courts of appeals are divided

over what judicial standards control

when the government seeks dismissal

under the state secrets privilege.

Even if the Court were to conclude that state

secrets dismissals are sometimes permitted under

Reynolds where a plaintiff can make its case without

privileged information, the courts below are

intractably divided about what the government must

show to obtain such a dismissal. These longstanding

divisions will remain unless this Court intervenes.

The circuits are divided over

whether the government may

obtain a state secrets dismissal

without establishing that the

privileged evidence supports a

“legally meritorious” defense.

The decision below—which dismissed Petitioner’s

lawsuit based on the government’s assertion of an

entirely hypothetical defense, and without any

judicial examination of the assertedly privileged

evidence—conflicts with the decisions of the D.C.,

Ninth, Sixth, and Second Circuits. Those circuits

require the government to establish, through ex parte,

in camera review, that the secret evidence would

provide a “valid” or “legally meritorious” defense—not

merely to hypothesize in the abstract about a defense

that might or might not have a factual and legal basis,

as the Fourth Circuit permitted here.

The Fourth Circuit dismissed the case based on the

government’s assertion that it might have defenses

that would implicate secret evidence. See App. 55–

58a, 62a. As the dissent explained, the government

“premise[d] its only defenses on far-fetched

hypotheticals”—including a claim that the NSA could,

in theory, use a system that deliberately avoided every

one

Petitioner’s

trillions

Internet

communications. App. 62a, 65–66a. Although the

court found that Petitioner had successfully put

forward a prima facie case, it granted dismissal

because it believed that any possible defenses would

involve privileged information. App. 55–58a. The

court did not examine any of the government’s

assertedly privileged evidence to see whether it in fact

supported a valid defense. Thus, even if the

government has no valid defense, and even though

Petitioner can make its case without any privileged

evidence, the court below dismissed this case

challenging ongoing and widespread constitutional

violations.

The Fourth Circuit has applied similar logic in

prior cases. Abilt, 848 F.3d at 316 (dismissing

employment discrimination claims against CIA);

Sterling v. Tenet, 416 F.3d 338, 347 (4th Cir. 2005)

(same); El-Masri, 479 F.3d at 309–10 (dismissing

plaintiff’s claims that he was abducted, imprisoned,

and tortured by the CIA because the court could

imagine possible defenses that would require secret

information).

Under the Fourth Circuit’s test, the government

need not present any defense at all, even in an ex

parte and in camera filing. The court need not review

the evidence that would underlie that defense to

determine whether it exists and is properly privileged.

And the court need not determine whether the defense

is valid. It is enough that the court can envision

“possible defenses” that would rely on sensitive

information. Id. at 310.

This approach conflicts with that of every other

circuit that has addressed how courts should evaluate

privileged defenses under Reynolds. These courts

have held that the government must show that it

would have a legally meritorious defense—one that

“would require judgment for the defendant.” In re

Sealed Case, 494 F.3d 139, 149–50 (D.C. Cir. 2007)

(reversing district court dismissal based on “possible

defenses”); see Molerio, 749 F.2d at 825; Jeppesen, 614

F.3d at 1083; Fazaga v. FBI, 965 F.3d 1015, 1067 (9th

Cir. 2020), rev’d in part on other grounds, 142 S. Ct.

1051 (2022); Tenenbaum, 372 F.3d at 777;

Zuckerbraun v. Gen. Dynamics Corp., 935 F.2d 544,

547 (2d Cir. 1991); cf. Bareford v. Gen. Dynamics

Corp., 973 F.2d 1138 (5th Cir. 1992).

Thus, all other courts to address the issue require

the government’s secret defense to be meritorious, not

merely possible. See In re Sealed Case, 494 F.3d at

149–50 (defining “valid defense”); Fazaga, 965 F.3d at

1067; Kasza, 133 F.3d at 1166; Tenenbaum, 372 F.3d

at 777–78. And they all require that determination to

be based on an “appropriately tailored in camera

review of the privileged record.” In re Sealed Case, 494

F.3d at 151; see Fazaga, 965 F.3d at 1067; Molerio, 749

F.2d at 825 (granting dismissal only after reviewing

evidence in camera); Tenenbaum, 372 F.3d at 777–78

(same).

As the D.C. Circuit has explained, “allowing the

mere prospect of a privileged defense to thwart a

citizen’s efforts to vindicate his or her constitutional

rights would run afoul of the Supreme Court’s caution

against precluding review of constitutional claims, . . .

and against broadly interpreting evidentiary

privileges[.]” In re Sealed Case, 494 F.3d at 151 (citing

Webster v. Doe, 486 U.S. 592, 603–04 (1988), and

United States v. Nixon, 418 U.S. 683, 710 (1974)).

“Were the valid-defense exception expanded to

mandate dismissal of a complaint for any plausible or

colorable defense, then virtually every case in which

the United States successfully invokes the state

secrets privilege would need to be dismissed.” Id. at

149–50.

Under the Fourth Circuit’s approach, even where

a plaintiff can establish a prima facie case that its

constitutional rights are being violated, and the

government has no valid defense, the court must

dismiss a challenge if a hypothetical defense might

implicate state secrets. That approach runs afoul of

Reynolds and the judiciary’s responsibility to decide

cases on the basis of evidence, rather than based on “a

system of conjecture.” In re Sealed Case, 494 F.3d at

150; see Reynolds, 345 U.S. at 9–10.

As the D.C. Circuit has acknowledged, the statesecrets dismissal remedy created by the lower federal

courts is an aberration in the law of evidence: it is a

“limited” exception to the longstanding principle that

successful invocation of the Reynolds privilege simply

results in the exclusion of certain sensitive evidence,

and that the case goes on without it—a rule favoring

neither plaintiffs nor defendants. In re Sealed Case,

494 F.3d at 148; see Part II supra. The rationale for

this valid-defense exception is that it would be a

“mockery of justice” for the court to allow a jury to

reach a manifestly “erroneous conclusion” when the

government has a meritorious defense but the need for

secrecy precludes public disclosure. Molerio, 749 F.2d

at 825; see In re Sealed Case, 494 F.3d at 148–51. But

there is no mockery of justice and no unfairness where

the government’s secret defense is not valid, or worse

yet, merely hypothetical, as here. See In re Sealed

Case, 494 F.3d at 150; In re United States, 872 F.2d at

481–82 (Ginsburg, J., concurring in part).

In an effort to minimize an obvious conflict with

the D.C. Circuit, the court below cited dicta from In re

Sealed Case, arguing that the D.C. Circuit’s validdefense rule does not apply where a court believes that

“any” conceivable defense would implicate privileged

information. App. 56a (quoting 494 F.3d at 149). But

that was not the D.C. Circuit’s holding. The court in

In re Sealed Case was simply describing the Fourth

Circuit’s ruling in El-Masri, not embracing it. Under

the D.C. Circuit’s reasoning, it does not matter

whether the government has one or many

hypothetical defenses; those defenses entitle it to

dismissal only if the relevant evidence exists and

supports a legally meritorious defense. See In re

Sealed Case, 494 F.3d at 150.

There is broad confusion in the

lower courts about the other

circumstances in which dismissal is

available under Reynolds.

The Fourth Circuit’s opinion also reflects broader

confusion in the lower courts about the circumstances

in which dismissal is available. Citing the

government’s hypothetical defenses, the court pointed

to a related basis for dismissing the case under the

privilege: where a court deems state secrets “so

central” to a case that proceeding would “present an

unjustifiable risk of disclosure.” App. 56–57a. This

test, however, is too amorphous to serve as a basis for

consistent judicial decision-making, and it has no

place where only the government’s defense could put

privileged information at issue. The court’s use of it

here shows why. Petitioner possesses no privileged

information and thus poses no risk of disclosing it.

Because Petitioner can establish its case using public

evidence, state secrets would be central only if the

government itself chooses to use privileged evidence

and that evidence establishes a defense. Yet rather

than assess whether the government’s hypothetical

defenses had any validity, see App. 64–66a, the Fourth

Circuit simply accepted the claim that privileged

information would be “so central” to the case and

dismissed.

Other courts have relied on different, but equally

amorphous tests to dismiss cases in which the plaintiff

could prevail using only public evidence. A number of

courts, for example, have held that a case may be

dismissed if the “very subject matter” of the suit is a

state secret. See, e.g., Kasza, 133 F.3d at 1166;

Zuckerbraun, 935 F.2d at 547–48. As noted above,

that erroneously conflates the Reynolds evidentiary

privilege with the Totten justiciability bar. See Part II

supra. But in addition, these courts do not explain how

to assess whether the “very subject matter” of a case

is a state secret, nor what criteria apply when a

plaintiff has unprivileged evidence that would

establish its prima facie case. Cf. Jeppesen, 614 F.3d

at 1084–85 (avoiding “the difficult question of

precisely which claims may be barred” under the “very

subject matter” test).

In other cases, courts have applied various

standards to dismiss suits based on a prediction about

the role that secret evidence would play in the

litigation. As noted above, the Fourth Circuit will

order dismissal if it predicts state secrets would be “so

central” to the litigation that the suit could not be

litigated without threatening the disclosure of secret

information. See, e.g., Abilt, 848 F.3d at 313; El-Masri,

479 F.3d at 306. The Ninth Circuit, in contrast,

permits dismissal where it deems privileged and

nonprivileged evidence “inseparable,” such that

“litigating the case to a judgment on the merits would

present an unacceptable risk of disclosing state

secrets.” Jeppesen, 614 F.3d at 1083; see also Bareford,

973 F.2d at 1144 (similar).

Some courts have eschewed such vague predictions

and instead require the development of evidence

through discovery before determining the role of state

secrets in the case. See, e.g., In re Sealed Case, 494

F.3d at 151 (declining to dismiss on state-secret

subject-matter grounds, noting that “the court has not

looked favorably upon broad assertions by the United

States that certain subject matters are off-limits for

judicial review”); Monarch Assurance P.L.C. v. United

States, 244 F.3d 1356, 1364 (Fed. Cir. 2001) (reversing

premature dismissal so that plaintiff could engage in

further discovery to support claim with nonprivileged

evidence); In re United States, 872 F.2d 472 at 477

(D.C. Cir. 1989) (refusing to dismiss on basis of the

government’s “unilateral assertion that privileged

information lies at the core of th[e] case”).

The lower courts also differ concerning whether

and when a court must consider alternatives to

dismissing a case on the basis of the privilege. Some

courts have dismissed on the basis of the privilege

without considering any alternatives. See, e.g.,

Tenenbaum, 372 F.3d at 777; Farnsworth Cannon,

Inc. v. Grimes, 635 F.2d 268, 281 (4th Cir. 1980).

Others have refused to dismiss where procedural

safeguards might enable the case to proceed. See, e.g.,

In re United States, 872 F.2d at 478 (discussing

measures to protect sensitive information as case

proceeds); see also Halpern v. United States, 258 F.2d

36, 41 (2d Cir. 1958) (refusing to dismiss suit because

case could be tried in camera).

Some courts have correctly held that where

dismissal might result from a successful invocation of

the privilege, the court must examine the actual

evidence over which the government has invoked the

privilege. See, e.g., Ellsberg v. Mitchell, 709 F.2d 51 at

59 n.37 (D.C. Cir. 1983); ACLU v. Brown, 619 F.2d

1170, 1173 (7th Cir. 1980). Others have declined to

examine the allegedly privileged evidence, relying

solely on secret affidavits submitted by the

government. See, e.g., Sterling, 416 F.3d at 344; Black

v. United States, 62 F.3d 1115, 1117–19 (8th Cir.

1995); Kasza, 133 F.3d at 1170.

Given this confusion, if dismissals are permitted at

all under Reynolds, it is vital that this Court clarify

that dismissal of a suit on the basis of the state secrets

privilege is appropriate solely when the removal of

privileged evidence renders it impossible for the

plaintiff to make out a prima facie case, or for the

defendant to present a valid defense.

IV.

This case is an ideal vehicle to decide

whether state secrets dismissals are

permitted under Reynolds and, if they

are, what standards control.

The case squarely presents the

dismissal question.

This case is an ideal vehicle to address state

secrets dismissals.

First, the procedural posture and factual showing

in Petitioner’s case squarely present the question of

whether a state secrets dismissal is ever proper where

a plaintiff can make out its case on nonprivileged

evidence. The court below held that Wikimedia had

put forward sufficient evidence that it was subjected

to Upstream surveillance, App. 24–35a, and therefore

the legality of that surveillance is properly presented

on the basis of nonprivileged information. Plaintiffs in

national security surveillance cases often lack

sufficient evidence to make a prima facie showing,

because absent some governmental disclosure, they

generally cannot show they were subject to the

program. See, e.g., ACLU v. NSA, 493 F.3d 644, 653

(6th Cir. 2007). In such cases, exclusion of the

privileged evidence causes plaintiffs to be unable to

meet their initial burden. As a result, the dismissal

issue only properly arises in cases where the plaintiff

successfully makes a prima facie showing using

unprivileged evidence. Wikimedia has done so here.

App. 22–35a.

Second, the record also squarely and specifically

raises the second question presented: namely,

whether, if state secrets privilege dismissals are ever

appropriate, the government must establish a valid

defense. Here, the court below dismissed without

assessing whether the government’s hypothetical

defense had any basis in fact, or was meritorious as a

legal matter. App. 55–58a; 62–66a.

Third, the Court can resolve these important

threshold questions without itself considering

classified information. If Petitioner is correct that

dismissal is unavailable or that the government

cannot prevail merely by asserting a hypothetical

defense, then the Court can remand for the lower

court to consider any privileged information as

necessary: for example, to conduct the “tailored in

camera review” necessary to determine whether

evidence supporting the government’s possible

defense exists and whether that evidence would

establish a valid defense. App. 62–66a.

The Fourth Circuit’s decision was

wrong.

The panel was wrong to dismiss Wikimedia’s suit

on state secrets grounds.

First, as explained in Part II supra, dismissal was

improper because the government invoked only the

Reynolds evidentiary privilege, not the distinct Totten

bar. The Reynolds evidentiary privilege does not

support dismissal where a plaintiff can proceed with

nonprivileged evidence.

Second, even if this Court were to hold that state

secrets dismissals are permitted under Reynolds, the

panel erred in concluding that the mere possibility of

a secret defense mandates dismissal. By refusing to

require in camera review to assess the validity of the

government’s defense, the panel deviated from the

valid-defense test in the D.C., Ninth, Sixth, and

Second Circuits, see Part III.A supra, and from this

Court’s admonition in Reynolds that “[j]udicial control

over the evidence in a case cannot be abdicated to the

caprice of executive officers.” 345 U.S. at 9–10.

As Judge Motz correctly observed in dissent,

“particularly when constitutional rights are at stake,

courts routinely probe a claim of privilege through an

‘appropriately tailored in camera review’ to determine

whether ‘resort to privileged material’ is in fact

necessary for the Government to pursue a ‘meritorious

and not merely plausible’ defense.” App. 64a (quoting

In re Sealed Case, 494 F.3d at 149–51). At a minimum,

the Fourth Circuit should have required such a review

here. 5

The Fourth Circuit’s assertion that secret evidence

is “so central” to the litigation that the privilege

requires dismissal does not justify its failure to assess

whether the government actually has any valid

defense. Where, as here, a plaintiff can proceed solely

on the basis of nonprivileged evidence, state secrets

could only be “central” if the government affirmatively

chooses to rely on secret evidence to defend itself—and

that scenario is specifically addressed by ex parte, in

camera review under the valid-defense test. If state

secrets dismissals are permitted under Reynolds, they

should be limited to cases in which the government

can in fact establish a meritorious defense.

Lastly, the government was wrong when it

suggested below that any ruling on the merits of

Petitioner’s claim would disclose state secrets. On

remand, the district court could assess—in an in

camera

proceeding—any

privileged

evidence

supporting any defense claimed by the government.

See App. 65a (“[T]he Government offers no reason why

an ‘appropriately tailored in camera review’ could not

ascertain the validity of the defense without

imperiling state secrets.” (quoting In re Sealed Case,

494 F.3d at 151)). If the court rejected the

government’s arguments after in camera review, its

public ruling would simply leave the existing public

evidence in place, and the case would proceed. If,

5 In addition, as Judge Motz acknowledged, allowing

Wikimedia’s case to proceed would not deprive the government

of the ability to mount a defense. App. 65–66a. Rather, the

government would remain free to dispute Wikimedia’s public

evidence by relying on its own public evidence and arguments,

just as it did at summary judgment. See id.

instead, it dismissed the case, its ruling would

indicate only that the government had established

some valid defense behind closed doors. It would not

reveal how or why the government had prevailed. See

Molerio, 749 F.2d at 825 (granting valid-defense

dismissal without revealing the basis for the defense).

The government should not be able to use the

theoretical possibility of a secret defense to foreclose

any judicial review of constitutional claims. Crediting

the

government’s

theories,

without

judicial

consideration of evidence, would surrender control of

a case to the executive branch—an approach to the

privilege that Reynolds forbids.

CONCLUSION

For the foregoing reasons, the petition for a writ of

certiorari should be granted.

Respectfully submitted,

Alex Abdo

Jameel Jaffer

KNIGHT FIRST AMENDMENT

INSTITUTE AT COLUMBIA

UNIVERSITY

475 Riverside Drive

Suite 302

New York, NY 10115

Patrick Toomey

Counsel of Record

Ashley Gorski

Sarah Taitz

AMERICAN CIVIL LIBERTIES

UNION FOUNDATION

125 Broad Street

18th Floor

New York, NY 10004

(212) 519-7816

ptoomey@aclu.org

Benjamin H. Kleine

Aarti Reddy

Maximilian Sladek de la Cal

David D. Cole

COOLEY LLP

3 Embarcadero Center

AMERICAN CIVIL LIBERTIES

20th Floor

UNION FOUNDATION

San Francisco, CA 94111

915 15th Street, NW

Washington, DC 20005

Deborah A. Jeon

David R. Rocah

AMERICAN CIVIL LIBERTIES

UNION FOUNDATION OF

MARYLAND

3600 Clipper Mill Rd., #350

Baltimore, MD 21211

Date: August 26, 2022

APPENDIX

APPENDIX A

PUBLISHED

UNITED STATES COURT OF APPEALS FOR THE

FOURTH CIRCUIT

No. 20-1191

WIKIMEDIA FOUNDATION,

Plaintiff−Appellant,

and

NATIONAL ASSOCIATION OF CRIMINAL

DEFENSE ATTORNEYS; HUMAN RIGHTS

WATCH; PEN AMERICAN CENTER; GLOBAL

FUND FOR WOMEN; THE NATION MAGAZINE;

THE RUTHERFORD INSTITUTE; WASHINGTON

OFFICE ON LATIN AMERICA; AMNESTY

INTERNATIONAL USA,

Plaintiffs,

NATIONAL SECURITY AGENCY/CENTRAL

SECURITY SERVICE; GENERAL PAUL M.

NAKASONE, in his official capacity as Director of

the National Security Agency and Chief of the

Central Security Service; OFFICE OF THE

DIRECTOR OF NATIONAL INTELLIGENCE;

RICHARD GRENELL, in his official capacity as

acting Director of National Intelligence; MERRICK

B. GARLAND, Attorney General; DEPARTMENT

OF JUSTICE,

Defendants–Appellees.

_________________________

CENTER FOR DEMOCRACY & TECHNOLOGY;

NEW

AMERICA’S

OPEN

TECHNOLOGY

INSTITUTE; DAVID H. KAYE, Evidence Law

Professor; EDWARD J. IMWINKELRIED, Evidence

Law Professor; D. MICHAEL RISINGER, Evidence

Law Professor; REBECCA WEXLER, Evidence Law

Professor; PROFESSOR STEPHEN I. VLADECK;

AMERICANS FOR PROSPERITY FOUNDATION;

BRENNAN CENTER FOR JUSTICE; ELECTRONIC

FRONTIER

FOUNDATION;

ELECTRONIC

PRIVACY

INFORMATION

CENTER;

FREEDOMWORKS

FOUNDATION;

TECHFREEDOM; NETWORK ENGINEERS AND

TECHNOLOGISTS,

Amici Supporting Appellant.

Appeal from the United States District Court of

Maryland, at Baltimore. T. S. Ellis, III, Senior District

Judge. (1:15−cv−00662−TSE)

_________________________

Argued: March 12, 2021 Decided: September 15, 2021

_________________________

Before MOTZ, DIAZ, and RUSHING, Circuit Judges.

_________________________

Affirmed by published opinion. Judge Diaz wrote the

majority opinion, in which Judge Motz joined as to

Parts I and II.A, and in which Judge Rushing joined

as to Part II.B.2 and C. Judge Motz wrote an opinion

concurring in part and dissenting in part. Judge

Rushing wrote an opinion concurring in part and in

the judgment.

_________________________

ARGUED: Patrick Christopher Toomey, AMERICAN

CIVIL LIBERTIES UNION FOUNDATION, New

York, New York, for Appellant. Joseph Forrest Busa,

UNITED STATES DEPARTMENT OF JUSTICE,

Washington, D.C., for Appellees. ON BRIEF:

Deborah A. Jeon, David R. Rocah, AMERICAN CIVIL

LIBERTIES

UNION

FOUNDATION

MARYLAND, Baltimore, Maryland; Ashley Gorski,

Charles Hogle, AMERICAN CIVIL LIBERTIES

UNION FOUNDATION, New York, New York;

Benjamin H. Kleine, COOLEY LLP, San Francisco,

California; Alex Abdo, Jameel Jaffer, KNIGHT FIRST

AMENDMENT

INSTITUTE

COLUMBIA

UNIVERSITY, New York, New York, for Appellant.

Ethan P. Davis, Acting Assistant Attorney General,

H. Thomas Byron III, Civil Division, UNITED

STATES DEPARTMENT OF JUSTICE, Washington,

D.C., for Appellees. Avery W. Gardiner, Gregory T.

Nojeim, Mana Azarmi, Stan Adams, CENTER FOR

DEMOCRACY & TECHNOLOGY, Washington, D.C.;

Sharon Bradford Franklin, Ross Schulman, NEW

AMERICA’S OPEN TECHNOLOGY INSTITUTE,

Washington, D.C.; Andrew A. Bank, Bret S. Cohen,

Allison M. Holt Ryan, Stevie N. DeGroff, HOGAN

LOVELS US LLP, Washington, D.C., for Amici Center

for Democracy & Technology and New America’s Open

Technology Institute. Benjamin B. Au, W. Henry

Huttinger, Los Angeles, California, Aditya V.

Kamdar, DURIE TANGRI LLP, San Francisco,

California, for Amici Evidence Law Professors.

Lauren Gallo White, San Francisco, California, Brian

M. Willen, WILSON SONSINI GOODRICH &

ROSATI PROFESSIONAL CORPORATION, New

York, New York, for Amicus Professor Stephen I.

Vladeck. Eric R. Bolinder, AMERICANS FOR

PROSPERITY FOUNDATION, Arlington, Virginia;

Sophia Cope, Mark Rumold, Andrew Cocker,

ELECTRONIC FRONTIER FOUNDATION, San

Francisco, California, for Amici Americans for

Prosperity Foundation, Brennan Center for Justice,

Electronic Frontier Foundation, Electronic Privacy

Information Center, FreedomWorks Foundation, and

TechFreedom. Jonathan Blavin, Elizabeth Kim,

Alexander Gorin, MUNGER, TOLLES & OLSON

LLP, San Francisco, California, for Amici Network

Engineers and Technologists.

_________________________

DIAZ, Circuit Judge:

We consider, for the second time, the Wikimedia

Foundation’s contentions that the government is

spying on its communications using Upstream, an

electronic surveillance program run by the National

Security Agency (“NSA”). In the first appeal, we found

Wikimedia’s allegations of Article III standing

sufficient to survive a motion to dismiss and vacated

the district court’s judgment to the contrary. On

remand, the court again dismissed the case, holding

that Wikimedia didn’t establish a genuine issue of

material fact as to standing and that further litigation

would unjustifiably risk the disclosure of state secrets.

Although the district court erred in granting

summary judgment to the government as to

Wikimedia’s standing, we agree that the state secrets

privilege requires the termination of this suit. We

thus affirm.

Our prior opinion contains many of the relevant

facts, including descriptions of the Upstream

surveillance program and its authorizing statute,

Section 702 of the Foreign Intelligence Surveillance

Act (“FISA”), 50 U.S.C. § 1881a. See Wikimedia

Found. v. Nat’l Sec. Agency/Cent. Sec. Serv., 857 F.3d

193, 200–07 (4th Cir. 2017). We take a moment here

to briefly review the inner workings of Upstream,

recap our previous decision, and relate what has

occurred since then.

As its name suggests, Upstream surveillance

involves the NSA’s collection of communications on

the Internet backbone, “upstream” of the Internet

user,

compelling

the

assistance

telecommunications-service providers. By contrast,

the NSA obtains the “vast majority” of Internet

communications collected under Section 702 directly

from a user’s Internet-service provider through the

PRISM surveillance program, Redacted, 2011 WL

10945618, at *9 & n.23 (FISA Ct. Oct. 3, 2011), which

isn’t at issue here.

The Internet backbone consists of domestic “highspeed, ultra-high bandwidth data- transmission lines”

and the relatively limited number of submarine and

terrestrial

circuits

that

carry

Internet

communications into and out of the United States,

J.A. 2739, which are often referred to as “chokepoint”

cables. More specifically:

The NSA performs Upstream surveillance by

first identifying a target and then identifying

“selectors” for that target. Selectors are the

specific means by which the target

communicates, such as e-mail addresses or

telephone numbers. Selectors cannot be

keywords (e.g., “bomb”) or names of targeted

individuals (e.g., “Bin Laden”).

The NSA then “tasks” selectors for collection

and sends them to telecommunicationsservice providers. Those providers must assist

the

government

intercepting

communications to, from, or “about” the

selectors. “About” communications are those

that contain a tasked selector in their content,

but are not to or from the target.

Wikimedia Found., 857 F.3d at 202. 1

Importantly, “[w]hile Upstream surveillance is

intended to acquire Internet communications, it does

so through the acquisition of Internet transactions.”

Id. at 203 (cleaned up). When an individual sends a

communication over the Internet, it’s broken up into

one or more data packets that are transmitted to, and

reassembled by, the receiving device. Each packet

travels separately across the Internet backbone. This

means that packets may take different paths to the

recipient, and while in transit, they’re mixed with

countless other packets making their own journeys.

“[A] complement of packets traversing the Internet

that together may be understood by a device on the

Internet” as one or many discrete communications

comprises an Internet “transaction.” Redacted, 2011

WL 10945618, at *9 n.23 (quoting a government

submission to the Foreign Intelligence Surveillance

Court (“FISC”)). “If a single discrete communication

within [a ‘multi-communication transaction’] is to,

from, or [until 2017] about, a Section 702-tasked

selector, and at least one end of the transaction is

foreign, the NSA will acquire the entire [multicommunication transaction].” Privacy & Civil

Liberties Oversight Board, Report on the Surveillance

Program Operated Pursuant to Section 702 of the

Foreign Intelligence Surveillance Act 44 (2014)

(“PCLOB Report”).

The NSA suspended its collection of “about”

communications in 2017 but continues to collect “to” and “from”

communications.

Wikimedia and eight other plaintiffs sued the

government, seeking “among other things, a

declaration that Upstream surveillance violates the

First and Fourth Amendments, an order permanently

enjoining the NSA from conducting Upstream

surveillance, and an order directing the NSA to purge

all records of Plaintiffs’ communications” obtained

through Upstream surveillance. Wikimedia Found.,

857 F.3d at 202 (cleaned up). The district court

dismissed the case for lack of Article III standing, and

the plaintiffs appealed.

Article III “[s]tanding is part and parcel of the

constitutional mandate that the judicial power of the

United States extend only to ‘cases’ and

‘controversies.’” Libertarian Party of Va. v. Judd, 718

F.3d 308, 313 (4th Cir. 2013) (quoting U.S. Const. art.

III, § 2). To establish standing, a plaintiff must show:

“(1) an injury in fact, (2) a sufficient causal connection

between the injury and the conduct complained of,

and (3) a likelihood that the injury will be redressed

by a favorable decision.” Susan B. Anthony List v.

Driehaus, 573 U.S. 149, 157–58 (2014) (cleaned up).

In what we called the “Wikimedia Allegation,”

Wikimedia claimed it had standing because (1) its

communications travel across every international

Internet link 2; (2) the NSA conducts Upstream

surveillance on at least one such link; and (3) “in order

for the NSA to reliably obtain communications to,

from, or about its targets in the way it has described,

the government must be copying and reviewing all the

Wikimedia uses “international” to describe something

occurring between the United States and a foreign country and

“international Internet link” to mean a chokepoint cable.

international text-based communications that travel

across a given link.” J.A. 57.

Together, these assertions were “sufficient to make

plausible the conclusion that the NSA is intercepting,

copying, and reviewing at least some of Wikimedia’s

communications,” establishing an injury-in-fact for a

Fourth Amendment violation. Wikimedia Found., 857

F.3d at 211. “And, because Wikimedia has selfcensored its speech and sometimes forgone electronic

communications” as a result of that surveillance, it

established an injury-in-fact for purposes of its First

Amendment claim. Id. Wikimedia also met the two

other requirements for standing because “Upstream

surveillance is the direct cause of the alleged injury,

and there’s no reason to doubt that the requested

injunctive and declaratory relief would redress the

harm.” Id. at 210.

We thus vacated the district court’s judgment as to

Wikimedia. We affirmed as to the other eight

plaintiffs, who alleged that given the government’s

incentives to cast a wide net, “the NSA is intercepting,

copying, and reviewing substantially all text-based

communications entering and leaving the United

States, including their own.” Wikimedia Found., 857

F.3d at 202 (cleaned up). We concluded that such

claims “about what the NSA ‘must’ be doing” based on

its goals “lack sufficient factual support to get ‘across

the line from conceivable to plausible.’” Id. at 214

(quoting Bell Atl. Corp. v. Twombly, 550 U.S. 544, 570

(2007)).

On remand, the district court ordered

jurisdictional discovery. But when Wikimedia sought

evidence related to Upstream, the NSA invoked the

state secrets privilege.

The privilege permits the United States to

“prevent the disclosure of information in a judicial

proceeding if ‘there is a reasonable danger’ that such

disclosure ‘will expose [matters of state] which, in the

interest of national security, should not be divulged.’”

El- Masri v. United States, 479 F.3d 296, 302 (4th Cir.

2007) (quoting United States v. Reynolds, 345 U.S. 1,

10 (1953)). In Reynolds, the decision that modernized

the privilege, three civilian observers aboard an Air

Force bomber testing secret electronic equipment died

when the plane caught fire and crashed. 345 U.S. at 3.

Their widows sued the United States under the

Federal Tort Claims Act and sought discovery related

to the incident. Id. at 3, 6. Instead of producing the

requested information, the Secretary of the Air Force

filed a formal claim of privilege, citing national

security concerns. Id. at 4–5. The Supreme Court

concluded that the government properly invoked the

privilege and sustained its refusal to disclose the

documents at issue. Id. at 6.

Thus, to invoke the state secrets privilege, the

United States must make “a formal claim of privilege,

lodged by the head of the department which has

control over the matter, after actual personal

consideration by that officer.” Id. at 7–8. Here, the

government filed the declaration of Daniel Coats, the

then Director of National Intelligence, who attested

that the disclosures requested by Wikimedia

“reasonably could be expected to cause serious

damage, and in many cases exceptionally grave

10a

damage, to the national security of the United

States.” 3 J.A. 174.

In particular, Director Coats asserted the

privilege over seven categories of information:

(A) information that would tend to confirm

what individuals or entities are subject to

Upstream

surveillance

activities;

(B)

information concerning the operational

details of the Upstream collection process; (C)

the

location(s)

which

Upstream

surveillance is conducted; (D) the categories of

Internet- based communications collected

through Upstream surveillance activities; (E)

information concerning the scope and scale of

Upstream surveillance; (F) NSA cryptanalytic

capabilities; and (G) additional categories of

classified information regarding Upstream

surveillance contained in opinions and orders

issued by, and submissions made to, the

[FISC].

J.A. 174–75.

Director Coats also confirmed several key facts

about Upstream surveillance. He explained that “in

the course of the Upstream collection process, certain

Internet transactions transiting the Internet

backbone

network(s)

certain

electronic

communication service provider(s) are filtered for the

purpose

excluding

wholly

domestic

communications.” J.A. 177. The NSA then scans the

remaining communications “to identify for acquisition

3 The government also filed the classified declaration of George

Barnes, the then Deputy Director of the NSA, describing the

national security concerns in greater detail.

11a

those transactions that are to or from” (or, until 2017,

“about”) the targeted selector and “ingest[s]” them

into government databases. J.A. 177–78.

Director Coats further acknowledged that the NSA

“is monitoring at least one circuit carrying

international Internet communications.” J.A. 186. But

he maintained that “[w]hile the Upstream collection

process has been described in general terms in this

declaration and in declassified documents and

unclassified reports, certain operational details of

Upstream collection remain highly classified.” J.A.

178.

Despite the NSA’s claim of privilege, Wikimedia

moved to compel discovery. Wikimedia argued that

FISA’s discovery procedures, as provided in 50 U.S.C.

§ 1806(f), displace the state secrets privilege in cases

involving government-run electronic surveillance.

This provision permits an “aggrieved person” who is

the target of electronic surveillance to request, under

certain circumstances, that the court conduct an in

camera and ex parte review of “the application, order,

and such other materials relating to the surveillance

as may be necessary to determine whether the

surveillance of the aggrieved person was lawfully

authorized and conducted.” 50 U.S.C. § 1806(f).

Wikimedia contended that (1) it had successfully

alleged that it was an aggrieved person; and (2) §

1806(f) required the district court to review evidence

related to Upstream surveillance in camera and ex

parte to determine whether the NSA lawfully

surveilled Wikimedia’s communications, instead of

dismissing the entire action.

The district court, however, concluded that FISA

doesn’t apply and denied Wikimedia’s motion. In

12a

particular, the court explained that the “§ 1806(f)

procedures do not apply where, as here, a plaintiff has

not yet established that it has been the subject of

electronic surveillance.” Wikimedia Found. v. Nat’l

Sec. Agency/Cent. Sec. Serv., 335 F. Supp. 3d 772, 780

(D. Md. 2018). Because Wikimedia had “merely

plausibly alleged that it has been the target of

surveillance and ha[d] not yet adduced evidence

establishing this fact of surveillance,” the court

determined that “it [wa]s not appropriate . . . to

engage in ex parte and in camera review of the

materials responsive to plaintiff’s interrogatories or to

those plaintiff describe[d] in its motion to compel.” Id.

at 786.

The government then moved for summary

judgment, contending that Wikimedia didn’t establish

a genuine issue of material fact as to the second or

third prongs of the Wikimedia Allegation 4 and that

the state secrets privilege independently requires

dismissal of the case. As we explain in further detail

below, the district court granted this motion, holding

that (1) Wikimedia established a genuine issue of

material fact as to the second but not the third prong

of the Wikimedia Allegation, 5 (2) the state secrets

The government didn’t dispute that Wikimedia had

established the first prong.

5 The district court held that Wikimedia “established” the

second prong of the Wikimedia Allegation “without a genuine

dispute as to any material fact.” Wikimedia Found. v. Nat’l Sec.

Agency/Cent. Sec. Serv., 427 F. Supp. 3d 582, 603 (D. Md. 2019).

Read literally, the court appears to have granted partial

summary judgment for Wikimedia on the second prong. But,

although Wikimedia opposed the government’s summary

judgment motion, it never filed its own motion. Nor did the court

13a

privilege foreclosed further litigation, and (3)

Wikimedia didn’t show any other injury that gives rise

to standing.

The district court first determined that Wikimedia

had established a genuine issue of material fact as to

the second prong of the Wikimedia Allegation, which

posits that the NSA conducts Upstream surveillance

on at least one international Internet link.

To prove this assertion, Wikimedia primarily

relied on a declassified 2011 FISC opinion, which

states that “the government readily concedes that

[the] NSA will acquire a wholly domestic ‘about’

communication if the transaction containing the

communication is routed through an international

Internet link being monitored by [the] NSA or is

routed through a foreign server.” Redacted, 2011 WL

10945618, at *15 (citing a government submission to

the FISC). An NSA witness confirmed the accuracy of

this sentence, and of the opinion generally, as of

October 2011. 6

But the government contended that the meaning

of the phrase “international Internet link” as used in

the FISC opinion isn’t the same as that used in the

invoke Federal Rule of Civil Procedure 56(f), which allows it to

grant summary judgment to a nonmovant under certain

circumstances. To square this circle, we assume that the district

court found only that Wikimedia established a genuine issue of

material fact on the second prong.

The district court recognized the quoted sentence as an

admissible statement by a party opponent. See Wikimedia

Found., 427 F. Supp. 3d at 602. On appeal, Wikimedia asserts,

and the government doesn’t dispute, that the NSA also adopted

the facts in the FISC opinion as a whole.

14a

Wikimedia Allegation. In fact, the NSA witness

testified that the “NSA has an understanding of this

term that is specific to how [the FISC] described it,”

but that its true definition can’t be confirmed or

denied because “it’s classified.” J.A. 447. And the

government pointed out that the 2011 FISC opinion

may not reflect the NSA’s current practices.

“Rather than belabor the squabble between the

parties about the meaning of this particular term” in

the FISC opinion, the district court zeroed in on an

entirely different government disclosure. Wikimedia

Found., 427 F. Supp. 3d at 602–03. The court sua

sponte relied on Director Coats’s statement that the

“NSA is monitoring at least one circuit carrying

international Internet communications” to conclude

that Wikimedia had produced sufficient evidence to

raise a genuine issue of material fact as to the second

prong of the Wikimedia Allegation (i.e., that the NSA

conducts Upstream surveillance on at least one

international Internet link). Id. at 603.

But the district court found that Wikimedia didn’t

establish a genuine issue of material fact as to the

third prong of the Wikimedia Allegation, which

asserts that the NSA is copying all communications on

a monitored link. At the motion-to-dismiss stage,

Wikimedia had alleged that “as a technical matter,

the government cannot know beforehand which

communications will contain selectors associated with

its targets, and therefore it must copy and review all

international text-based communications transiting a

circuit in order to identify those of interest.”

Wikimedia Found., 857 F.3d at 204 (cleaned up).

15a

To undermine that claim, the government offered

the declarations of Henning Schulzrinne, an “expert

in internet technology.” Appellee’s Br. at 44.

Schulzrinne wasn’t privy to any classified or other

non-public information about how the NSA actually

operates Upstream surveillance, so he instead opined

that the NSA could “in theory” use a technique called

traffic mirroring to conduct Upstream-style

surveillance

without

copying

Wikimedia’s

communications. J.A. 719.

According to Schulzrinne, traffic mirroring

requires installing a link (i.e., a fiber-optic cable)

between the surveilling entity’s equipment and a

mirror port on the router or switch directing Internet

traffic at the target location. The router or switch is

then configured to copy traffic from one link to another

without interrupting the original. It can also be

programmed to whitelist or blacklist certain IP

addresses, thereby filtering the data before copying it.

Whitelisting involves copying only communications

from specific IP addresses, while blacklisting involves

copying everything except communications from

specific IP addresses.

Wikimedia responded with the declarations of

Scott Bradner, “an Internet networking expert.”

Appellant’s Br. at 23. Although Bradner conceded that

it’s “technically possible” to use traffic mirroring with

filtering (as envisioned by Schulzrinne), “doing so

would purposefully ignore most of the Internet” and

“would be inconsistent with the publicly known

details about the [U]pstream collection program.” J.A.

3898.

Bradner explained that traffic mirroring with

filtering “would require either that the [Internet

16a

Service Provider (“ISP”)] agree to place an NSAoperated device into the heart of its network”—which

could negatively impact “the ISP’s network in the

event of an equipment failure or misconfiguration—or

that the ISP’s personnel have enough knowledge of

the filter criteria to configure the ISP’s router.” J.A.

1023. Moreover, these filters would “place potentially

significant additional demands on the router’s

processing power, which could affect the performance

of the router and create a risk of overloading the

router, thereby interfering with the ISP’s ability to

support its customers’ traffic.” J.A. 1025.

Bradner further opined that rather than traffic

mirroring with filtering, the NSA is “most likely”

using link-layer copying (essentially traffic mirroring

without filtering) or optical splitters. J.A. 1022. An

optical splitter is a physical device attached to a fiberoptic cable that reflects a portion of the light traveling

down that circuit to a different receiver. The

information continues on its original course, while an

exact duplicate is sent to the surveilling entity. Any

filtering must take place after the copy is made. The

technology is “extremely reliable as it consumes no

power, has no software, and cannot slow traffic.”

Technologists’ Amicus Br. at 10; see also J.A. 3921.

According to Bradner, link-layer copying and optical

splitters offer the NSA the “greatest operational

control and confidentiality in carrying out upstream

collection with the least risk of interference with the

ISP’s ordinary network operations.” J.A. 1025.

Bradner also pointed to several government

disclosures as evidence that the NSA is copying all

communications on a monitored link. These include

the previously discussed statement from the 2011

FISC opinion, which provides that the “NSA will

17a

acquire a wholly domestic ‘about’ communication if

the transaction containing the communication is

routed through an international Internet link being

monitored by [the] NSA or is routed through a foreign

server,” Redacted, 2011 WL 10945618, at *15; and a

government report stating that the NSA’s goal is to

“comprehensively acquire communications that are

sent to or from its targets,” PCLOB Report 10, 37,

123. 7

Of these disclosures, the district court mentioned

only the PCLOB Report in the context of Bradner’s

opinion that the NSA is likely conducting Upstream

surveillance using link-layer copying or optical

splitters rather than traffic mirroring with filtering.

See Wikimedia Found., 427 F. Supp. 3d at 603–10. It

declined to consider this opinion, reasoning that it

7 Wikimedia’s evidence also included (1) another government

report revealing that the NSA had more than 120,000 Section

702 targets in 2017, Office of the Director of National Intelligence

Statistical Transparency Report for 2017 (Apr. 2018); (2) “[t]he

leading treatise on national security investigations, co-authored

by the former Assistant Attorney General for National Security,”

Appellant’s Br. at 31–32 (citing David Kris & J. Douglas Wilson,

Nat’l Security Investigations & Prosecutions 2d § 17.5 (2015));

(3) “[r]ecent disclosures by the United Kingdom about

functionally equivalent surveillance undertaken by the NSA’s

British counterpart,” id. at 32 (citing Further Observations of the

Government of the United Kingdom, Human Rights

Organizations v. United Kingdom, Eur. Ct. H.R. (Dec. 16, 2016),

https://privacyinternational.org/sites/default/files/201802/2016.12.16%20Government%27s%20further%20obs.pdf); and

(4) descriptions of “the U.S. government’s EINSTEIN 2

surveillance program, which protects government networks

through a similar form of Internet surveillance,” id. (citing Legal

Issues Relating to the Testing, Use, and Deployment of an

Intrusion-Detection System (EINSTEIN 2.0), 33 Op. O.L.C. 1

(Jan. 9, 2009)).

18a

rests on “speculative assumptions about the NSA’s

surveillance practices and priorities and [its]

resources and capabilities.” Id. at 604–05 (citing Fed.

R. of Evid. 702 and Daubert v. Merrell Dow Pharms.,

Inc., 509 U.S. 579 (1993)).

The court instead focused on the technical

arguments presented by both sides and concluded that

the record didn’t establish that the NSA must copy all

communications on a surveilled circuit by

“technological necessity.” Id. at 609. It therefore held

that Wikimedia had failed to establish a genuine issue

for trial as to standing.

The district court next “assum[ed] arguendo that[]

there is a genuine dispute of material fact as to the

third prong of the Wikimedia Allegation,” yet still

determined that the case must be dismissed because

of the state secrets privilege. Id. at 610.

In doing so, the court again rejected Wikimedia’s

argument that FISA displaces the state secrets

privilege in this case. This time, it distinguished

Wikimedia’s case from the only other circuit case

directly addressing this issue, Fazaga v. FBI, 916 F.3d

1202 (9th Cir. 2019), amended on denial of reh’g en

banc by 965 F.3d 1015 (9th Cir. 2020), cert. granted,

2021 WL 2301971 (June 7, 2021), which holds that

FISA’s discovery procedures in § 1806(f) apply instead

of the state secrets privilege under certain

circumstances.

The plaintiffs there challenged a counter-terrorism

investigation involving electronic surveillance

conducted by an informant for the Federal Bureau of

Investigation. “Several sources,” including the

19a

Bureau, had confirmed the identity of the informant

and that he “created audio and video recordings” for

the investigation. Fazaga, 965 F.3d at 1028.

The Fazaga district court dismissed all but one of

the plaintiffs’ claims at the pleading stage based on

the government’s assertion of the state secrets

privilege. But the Ninth Circuit reversed, holding that

FISA displaces the privilege whenever “an aggrieved

person affirmatively challenges, in any civil case, the

legality of electronic surveillance or its use in

litigation, whether the challenge is under FISA itself,

the Constitution, or any other law” and that “[t]he

complaint’s allegations are sufficient if proven to

establish that Plaintiffs are ‘aggrieved persons’” who

had been subjected to electronic surveillance by the

government. Id. at 1030, 1053.

The district court here determined that even if

Fazaga were binding in our circuit such that § 1806(f)

displaces the state secrets privilege, it wouldn’t help

Wikimedia. That’s because the Ninth Circuit found

the Fazaga plaintiffs’ allegations sufficient to

establish that they were aggrieved persons (as

required to apply § 1806(f)) at the motion-to-dismiss

stage. Wikimedia, on the other hand, faced summary

judgment and thus needed to establish a genuine

issue of material fact that it was the subject of

electronic surveillance but had failed to do so.

Accordingly, the court concluded once more that the §

1806(f) procedures don’t apply.

That being so, the court turned to the government’s

claim of privilege. It determined that state secrets are

“so central” to litigating the Wikimedia Allegation

that “the defendants cannot properly defend

themselves without using privileged evidence,”

20a

Wikimedia Found., 427 F. Supp. 3d at 613, and that

further proceedings “would present an unjustifiable

risk” of disclosing privileged information, id. at 612.

The court thus ruled that the case must also be

dismissed because of the state secrets privilege.

Finally, the district court concluded that none of

Wikimedia’s other alleged injuries independently

establish standing. In addition to the Wikimedia

Allegation, Wikimedia asserted that: (1) Upstream

surveillance impaired Wikimedia’s communications

with its community members, as evidenced by the

drop in readership for certain Wikipedia pages; (2)

Wikimedia had to take costly protective measures

against Upstream surveillance; and (3) Wikimedia

has third party standing to assert its users’ rights.

The court held that the first two theories fail under

Clapper v. Amnesty International USA, 568 U.S. 398

(2013), and the last collapses under its own weight.

In Clapper, several plaintiffs in the United States

challenged Section 702, alleging that their work

“requires them to engage in sensitive international

communications with individuals who they believe are

likely targets of surveillance.” 568 U.S. at 401. As we

described in our prior opinion, the plaintiffs there

“had two separate theories of Article III standing: (1)

there was an ‘objectively reasonable likelihood’ that

their communications would be intercepted in the

future pursuant to Section 702 surveillance, and (2)

they were forced to undertake costly and burdensome

measures to avoid a substantial risk of surveillance.”

Wikimedia Found., 857 F.3d at 206 (quoting Clapper,

568 U.S. at 407). “They did not, however, have actual

knowledge of the Government’s Section 702 targeting

21a

practices.” Id. (cleaned up). The Supreme Court held

that neither theory was sufficient to prove standing at

the summary-judgment stage because they depended

on a “speculative chain of possibilities [that] does not

establish that injury based on potential future

surveillance is certainly impending or is fairly

traceable” to Section 702 surveillance. Clapper, 568

U.S. at 414.

As relevant to Wikimedia’s claim of decreased

readership, Clapper explained that “a chilling effect

arising merely from the individual’s knowledge that a

governmental agency was engaged in certain

activities or from the individual’s concomitant fear

that, armed with the fruits of those activities, the

agency might in the future take some other and

additional action detrimental to that individual”

doesn’t establish standing. 568 U.S. at 417–18; see

also id. at 418 (“Because allegations of a subjective

‘chill’ are not an adequate substitute for a claim of

specific present objective harm or a threat of specific

future harm, the plaintiffs . . . lack standing.” (cleaned

up)).

The district court found that Wikimedia made a

similarly deficient assertion: that its “decreased

readership is a result of individual[] fear that the

government might be monitoring their Internet

activity and might use that information at some later

date.” Wikimedia Found., 427 F. Supp. 3d at 616. The

court then determined that Wikimedia otherwise

lacked objective evidence of “an ongoing and sustained

drop in [its] readership,” or that any such decline

stemmed from “Upstream surveillance specifically”

rather than “media coverage of NSA surveillance

generally.” Id. (cleaned up). It thus concluded that

22a

Wikimedia’s reduction in readership didn’t establish

standing.

In assessing Wikimedia’s standing based on

protective measures, the district court pointed to

Clapper’s admonition that plaintiffs “cannot

manufacture standing merely by inflicting harm on

themselves based on their fears of hypothetical future

harm.” 568 U.S. at 416. By finding the summary

judgment record inadequate to establish the

Wikimedia Allegation, the court had already ruled

that any harm to Wikimedia from Upstream

surveillance was “purely hypothetical” and thus

insufficient to prove standing. Wikimedia Found., 427

F. Supp. 3d at 617. As an additional nail in the coffin,

the court observed that Wikimedia began

implementing its protective measures years before

learning about Upstream for a variety of other

reasons, “including protecting against individual

computer hackers and keeping . . . company policies

up-to-date and transparent,” so the requested

injunctive and declaratory relief “would not redress

any

alleged

injury

from

these

protective

expenditures.” Id. at 617 n.63.

Nor was the district court persuaded that

Wikimedia had established third party standing. For

third party standing, a plaintiff must demonstrate “(1)

an injury-in-fact; (2) a close relationship between

[itself] and the person whose right [it] seeks to assert;

and (3) a hindrance to the third party’s ability to

protect his or her own interests.” Freilich v. Upper

Chesapeake Health Inc., 313 F.3d 205, 215 (4th Cir.

2002).

The court held that Wikimedia didn’t satisfy any of

these elements. As discussed, the court had already

23a

rejected all of Wikimedia’s alleged injuries-in-fact.

The court also determined that, unlike lawyers and

clients or doctors and patients, Wikimedia doesn’t

have the requisite “protected, close relationships”

with its “largely unidentified contributors.”

Wikimedia Found., 427 F. Supp. 3d at 617 & n.65. In

fact, Wikimedia “only presented declarations from one

single contributor,” who claimed that the “normal

burdens of litigation” and her “workload as a medical

student” make it “impossible” for her to bring suit. Id.

at 617–18. The court found this “insufficient” to show

that an obstacle prevents her from protecting her own

interests. Id. at 618.

The district court thus dispatched all of

Wikimedia’s theories of standing, dismissed the case,

and entered judgment for the government. This

appeal followed.

II.

Wikimedia contends that the district court erred in

dismissing its case because (1) the evidence it

presented establishes a genuine dispute of material

fact with respect to its standing; (2) FISA displaces the

state secrets privilege in this context; (3) even if FISA

doesn’t apply, the state secrets privilege doesn’t

require dismissal because “Wikimedia can establish

its standing without resort to privileged evidence[]”

and the government hasn’t shown that it can’t defend

itself without privileged evidence; and (4) “Wikimedia

has presented evidence of additional injuries” that

don’t implicate any state secrets. Appellant’s Br. at 14,

17.

As we explain, the record evidence is sufficient to

establish a genuine issue of material fact as to

Wikimedia’s standing. But FISA doesn’t displace the

24a

state secrets privilege, and further litigation would

unjustifiably risk the disclosure of privileged

information. And because Wikimedia’s other alleged

injuries don’t provide independent bases for standing,

this case must be dismissed.

Because standing is jurisdictional, we begin our

discussion there. See Libertarian Party of Va., 718

F.3d at 313. Our review of a district court’s decision

on summary judgment is de novo, and we view all

facts and reasonable inferences in the light most

favorable to the nonmovant—here, Wikimedia. See

Sylvia Dev. Corp. v. Calvert Cnty., 48 F.3d 810, 817

(4th Cir. 1995).

The government maintains that Wikimedia hasn’t

established a genuine issue for trial as to standing on

the second prong of the Wikimedia Allegation: that

Upstream surveillance occurs on at least one

international Internet link. We disagree.

Wikimedia contends that an “international

Internet link” is a “chokepoint” cable, which refers to

one of the relatively limited number of circuits that

carry Internet communications into and out of the

United States. Because Wikimedia claims that its

communications traverse all chokepoint cables—but

not necessarily all other circuits on the Internet

backbone—the second prong of the Wikimedia

Allegation centers on showing that the NSA is

monitoring at least one of these chokepoint cables.

The problem for Wikimedia is that the NSA never

uses the words “chokepoint cable” in its public

disclosures. The NSA does, however, use the phrase

25a

“international Internet link.” The parties therefore

dispute whether those terms are interchangeable, and

even if they are, whether the relevant disclosures

reveal that the NSA is actually monitoring such a

circuit.

At the outset, we focus on the government’s

concession in the 2011 FISC opinion and not the Coats

declaration (on which the district court relied). As

Wikimedia acknowledges, Director Coats’s statement

that the NSA is “monitoring at least one circuit

carrying international Internet communications,” J.A.

186, doesn’t identify where on the Internet backbone

that circuit is located. The Coats declaration thus

doesn’t show that the NSA is monitoring a chokepoint

cable.

By contrast, the FISC opinion recites a

government concession in that case “that [the] NSA

will acquire a wholly domestic ‘about’ communication

if the transaction containing the communication is

routed through an international Internet link being

monitored by [the] NSA.” Redacted, 2011 WL

10945618, at *15 (emphasis added). According to

Wikimedia, this concession—the accuracy of which

the NSA has confirmed in this case—is evidence that

NSA is in fact monitoring a circuit carrying

international communications.

The government argues that the statement from

the FISC opinion doesn’t reveal that the NSA is

actually monitoring an international Internet link.

Rather, it conveys only that if the NSA is monitoring

such a link, the agency will acquire the

communications traversing it. But the government’s

strained construction ignores grammar. The

consequence described in the independent clause (i.e.,

26a

the NSA’s acquisition of a domestic communication) is

tied to a conditional clause that turns on whether the

transaction is on an international Internet link that

the NSA is monitoring—not whether the NSA is

monitoring such a link at all. The sentence is thus

premised on the NSA surveilling at least one

international Internet link, over which a transaction

of interest may travel.

The government also says that Wikimedia has no

evidence that the NSA still adheres to these practices,

even if it did in 2011, and that “at least the conclusion

of this conditional statement is no longer accurate”

because “‘about’ collection ended in 2017.” Appellee’s

Br. at 40 n.3 (quoting Wikimedia Found., 427 F. Supp.

3d at 602 n.38). But the government never says that

the way it acquired “about” communications differs

from the way it collects “to” and “from”

communications. Nor have we seen anything in the

record to suggest that.

To the contrary, Upstream collection is often

described as a single process across all types of

communications. See, e.g., Redacted, 2011 WL

10945618, at *11 (“[The] NSA’s upstream Internet

collection devices are generally incapable of

distinguishing between transactions containing only a

single discrete communication to, from, or about a

tasked selector and transactions containing multiple

discrete communications, not all of which may be to,

from, or about a tasked selector.”). And given the lack

of evidence that the NSA has changed the way it

operates Upstream, it’s reasonable to infer that the

government’s concession in the FISC opinion remains

accurate despite the passage of time and even though

the government no longer retains “about”

communications.

27a

Next, the government argues that Wikimedia

didn’t dispute the district court’s determination that

“the differences between the term ‘international

internet link’ and the term [‘chokepoint cable’] . . .

cannot be known without violation of the state secrets

privilege.” Wikimedia Found., 427 F. Supp. 3d at 602.

Because the “opening brief does not mention the

district court’s ruling on this point, much less argue

that the court erred or explain how [it erred],” the

government contends that Wikimedia failed to

preserve this issue. Appellee’s Br. at 41 (citing

Grayson O Co. v. Agadir Int’l LLC, 856 F.3d 307, 316

(4th Cir. 2017)).

But it’s not clear that the district court actually

ruled on the definition of “international Internet link,”

as opposed to merely describing the government’s

position on it. See Wikimedia Found., 427 F. Supp. 3d

at 602 (“Defendants, however, assert …. Thus, the

differences between the term….”). The court then

turned away from the FISC opinion to focus on the

Coats declaration, suggesting that the court didn’t

intend to resolve the significance of “international

Internet link” at all—on state secrets or any other

grounds. Id. at 602–03.

Even if the court was commenting on the secret

nature of the phrase “international internet link,” it

ultimately found for Wikimedia on the second prong.

Wikimedia thus had no reason to raise any arguments

on the second prong before the government contested

it. 8 See Nw. Airlines, Inc. v. Cnty. of Kent, 510 U.S.

8 In any event, Wikimedia’s opening brief explains how the

FISC opinion supports the second prong based on publicly

available information. This preserves its arguments on the point.

See Appellant’s Br. at 25–27; see also Blackwelder v. Millman,

28a

355, 364 (1994) (“A prevailing party need not . . .

defend a judgment on any ground properly raised

below, so long as that party seeks to preserve, and not

to change, the judgment.”).

The government argues that the meaning of

“international Internet link” is in fact classified and

that Wikimedia thus lacks evidence showing that the

meaning of those words as used in the FISC opinion is

the same as that used in the Wikimedia Allegation.

But the government’s insistence that the true

definition of this phrase is a secret doesn’t invalidate

its concession in the FISC opinion as “concrete

evidence from which a reasonable juror could return a

verdict in [Wikimedia’s] favor.” Anderson v. Liberty

Lobby, Inc., 477 U.S. 242, 256 (1986). Even if

“international Internet link” could conceivably be code

for anything from a chihuahua to a chandelier, it’s

sensible to infer that the FISC opinion uses that

phrase to refer to a chokepoint cable. 9

Indeed, the ordinary meaning of “international

Internet link” is a connection carrying Internet traffic

between two countries. And its usage in describing

Upstream surveillance suggests that one of those

countries must be the United States. See, e.g., PCLOB

Report at 40 (“Upstream collection . . . [occurs] with

the compelled assistance (through a Section 702

522 F.2d 766, 771 (4th Cir. 1975) (holding that a prevailing party

“may support the judgment by urging any theory, argument, or

contention which is supported by the record, even though it was

specifically rejected by the lower court”).

9 Wikimedia also argues that it’s reasonable to infer from

government disclosures that the NSA is monitoring multiple

chokepoints. But this assertion is superfluous to what Wikimedia

must prove for standing, so we don’t address it further.

29a

directive) of the providers that control the

telecommunications backbone . . . . The collection

therefore does not occur at . . . foreign telephone or

Internet companies, which the government cannot

compel to comply with a Section 702 directive.”

(emphasis added)); id. at 36–37 (“Once tasked,

selectors used for the acquisition of upstream Internet

transactions are sent to a United States electronic

communication

service

provider

acquire

communications that are transiting through . . . the

‘Internet backbone.’” (emphasis added)); J.A. 1003

(Bradner Decl.) (“This of course makes sense, given

that public Internet traffic on [chokepoint cables] . . .

is the traffic that the NSA is authorized to monitor

under its Section 702 procedures.”).

To be sure, “whether an inference is reasonable

cannot be decided in a vacuum; it must be considered

in light of the competing inferences to the contrary.”

Sylvia Dev. Corp., 48 F.3d at 818 (cleaned up). But the

government doesn’t offer any evidence suggesting that

“international Internet link” has a counterintuitive

meaning.

Wikimedia’s

argument

that

the

government’s concession in the FISC opinion refers to

a chokepoint cable thus falls well “within the range of

reasonable probability.” Id. (quoting Ford Motor Co. v.

McDavid, 259 F.2d 261, 266 (4th Cir. 1958)). And

because we must draw all reasonable inferences in

Wikimedia’s favor, this is sufficient to establish a

genuine issue of material fact as to the second prong

of the Wikimedia Allegation.

Wikimedia next argues that summary judgment

for the government wasn’t appropriate as to the third

prong of the Wikimedia Allegation: that the NSA

30a

copies all communications on a monitored link. In

particular, Wikimedia asserts that this prong is

supported by (1) “the government’s own disclosures”;

(2) the “technical and practical necessities” of

conducting Upstream surveillance; and (3) the NSA’s

goal of “comprehensively acquir[ing] communications

that are sent to or from its targets.” Appellant’s Br. at

28–30. Relatedly, Wikimedia contends that the court

shouldn’t have excluded a portion of Bradner’s expert

opinion when assessing this prong.

We agree in part. Because reasonable inferences

drawn from the government’s concession in the FISC

opinion establish a genuine issue of material fact as to

the third prong, the district court erred in granting

summary judgment to the government.

The government doesn’t dispute that Wikimedia

may prove the third prong by showing that the NSA is

copying all transactions on a monitored link by choice,

as Wikimedia urges now, rather than by technological

necessity, as it argued at the motion- to-dismiss stage.

This shift in focus is hardly surprising, given

Wikimedia’s acknowledgment that it’s technically

feasible to conduct Upstream surveillance without

copying all communications on a monitored link.

The district court, when discussing the third prong

of the Wikimedia Allegation, made no mention of most

of the government disclosures Wikimedia cited for its

claim that the NSA is copying all communications

transiting a monitored link by choice. To the extent

that the court touched on copying by choice at all, it

did so only in the context of excluding from its analysis

Bradner’s expert opinion that discusses why the NSA

might prefer link-layer copying or optical splitters

(which both result in wholesale copying). We thus

31a

conduct the analysis that the district court passed on:

whether Wikimedia “set forth specific facts showing

that there is a genuine issue for trial” with respect to

the allegation that the NSA has elected to copy all

transactions on a surveilled circuit. Lujan v. Nat’l

Wildlife Fed., 497 U.S. 871, 888 (1990).

As support for this proposition, Wikimedia again

leads with the same statement from the 2011 FISC

opinion, this time highlighting a different portion of it.

The government’s concession in that case that the

“NSA will acquire a domestic ‘about’ communication if

the transaction containing the communication is

routed through an international Internet link being

monitored by [the] NSA” can only be true, says

Wikimedia, if the NSA is copying all traffic on a

surveilled circuit. 10 Redacted, 2011 WL 10945618, at

*15 (emphasis added).

The government says that this portion of the FISC

opinion lacks technical precision. In particular, it

points to another part of the FISC opinion that says

the

“NSA

may

acquire

wholly

domestic

communications,” id. at *11 n.34 (emphasis added).

which it claims is inconsistent with the “will acquire”

statement. This argument, however, takes the “may

acquire” quote out of context.

The “may acquire” phrase comes from a portion of

the opinion describing a specific kind of transaction (a

10 The “will acquire” language also appears once more in the

FISC opinion, expressing essentially the same notion. See

Redacted, 2011 WL 10945618, at *11 (“[The] NSA likely acquires

tens of thousands more wholly domestic communications every

year, given that [the] NSA’s upstream collection devices will

acquire a wholly domestic ‘about’ [communication] if it is routed

internationally.” (emphasis added)).

32a

multi-communication transaction), and not how

transactions on a monitored link are generally

acquired. One or more of the discrete communications

contained within a single multi-communication

transaction may be wholly domestic but the NSA may

“lack[] sufficient information . . . to determine the

location or identity” of the sender. Id. Accordingly,

“[the]

NSA

may

acquire

wholly

domestic

communications” within a particular multicommunication transaction without knowing that it

has done so. 11 Id. (emphasis added). But this says

nothing about how the NSA obtained the multicommunication

transaction—i.e.,

whether

it’s

because, as Wikimedia alleges, the NSA is copying all

transactions on a monitored link.

The government also offers a competing

interpretation of its concession in the FISC opinion. It

argues that the “will acquire” quote doesn’t mean the

NSA acquires every domestic communication on a

monitored link. Why? Because, posits the government,

the relevant sentence says only the NSA will acquire

“a” domestic communication, not “all” such

communications.

While

literally

true,

the

government’s myopic reading ignores the significance

of the word “a” in context.

As an indefinite article, “a” can mean “any” and

precedes a “singular noun[] when the referent is

unspecified.”

MERRIAM-WEBSTER,

https://www.merriam-webster. com/dictionary/a (last

In fact, when the NSA manually reviewed a random

sample of transactions collected through Upstream, it couldn’t

“determine conclusively” whether 224 out of 5,081 multicommunication transactions contained wholly domestic

communications. Redacted, 2011 WL 10945618, at *11 n.34.

33a

visited August 18, 2021). Therefore, the best reading

of the government’s concession is that the NSA “will

acquire”

any

single,

unspecified

domestic

communication, so long as it’s traversing a monitored

international Internet link. In the context of the “will

acquire” sentence then, the NSA’s surefire acquisition

of “a” domestic communication on a surveilled circuit

is equivalent to its acquisition of “all” such

transactions.

Judge Rushing says that we take the government’s

concession in the FISC opinion out of context. Not so.

The fact that the “will acquire” phrase appears in a

section of the FISC opinion explaining that the NSA

intentionally designed its collection devices to acquire

wholly domestic communications is entirely

consistent with the inference that the NSA has chosen

to copy all communications on a monitored link.

Moreover, the government’s concession isn’t a

stray statement swimming against a tide of contrary

text. In fact, only a few subparts of the eighty-page

FISC opinion are relevant to how the NSA acquires

transactions. See Redacted, 2011 WL 10945618, at *9–

16 (discussing the scope of the NSA’s Upstream

collections and the NSA’s targeting procedures). And

it’s telling that the FISC opinion recites the “will

acquire” language a second time, see id. at *11, when

describing the government’s collection of wholly

domestic communications in a portion of the opinion

dedicated to “the comprehensiveness of the NSA’s

collection practices,” Concurrence at 66. Indeed,

neither the government nor my colleague have

pointed to a single sentence in the other seventy-nine

34a

pages of the FISC opinion that refutes Wikimedia’s

interpretation of the government’s concession. 12

Wikimedia’s “grab-bag” of other support for the

third prong, Appellee’s Br. at 51, doesn’t contain

standalone proof that the NSA is copying before

filtering. For example, the NSA’s desire to be

“comprehensive[]” in its surveillance, PCLOB Report

at 10, 123, doesn’t necessarily mean that its collection

of communications is exhaustive, especially given the

agency’s technical, logistical, and financial restraints

in the face of competing mission priorities—all of

which are classified.

But Wikimedia’s supplemental evidence is at least

consistent with its reasonable interpretation of the

government’s concession in the FISC opinion, and the

government again fails to offer any contradictory

evidence that casts doubt on those inferences.

Wikimedia thus has established a genuine issue of

material fact with respect to the third prong of the

Wikimedia Allegation and its Article III standing. 13

At best, Judge Rushing’s belief that the government’s

concession in the FISC opinion can be reasonably interpreted

another way confirms that Wikimedia has raised a genuine issue

of material fact sufficient to preclude summary judgment. See W.

C. English, Inc. v. Rummel, Klepper & Kahl, LLP, 934 F.3d 398,

404 (4th Cir. 2019) (explaining that when there are “two

reasonable interpretations” of a phrase, “the granting of

summary judgment for either side [is] improper” (cleaned up)).

Because we hold that the case must nonetheless be

dismissed because of the state secrets privilege, we don’t tackle

Wikimedia’s claim that the district court abused its discretion in

excluding some of Bradner’s opinions. Nor do we address

35a

Having confirmed our jurisdiction, we now turn to

Wikimedia’s contention that the court erred in relying

on the state secrets privilege to deny its motion to

compel discovery and grant the government’s motion

to dismiss because § 1806(f) of FISA displaces the

privilege. 14 We review de novo both questions of

statutory interpretation, United States v. Abugala,

336 F.3d 277, 278 (4th Cir. 2003), and “legal

determinations involving state secrets,” El-Masri v.

United States, 479 F.3d 296, 302 (4th Cir. 2007).

Because we conclude that § 1806(f) is relevant only

when a litigant challenges the admissibility of the

government’s surveillance evidence, it doesn’t apply

here. Instead, we apply the state secrets privilege and

hold, like the district court, that it forecloses further

litigation.

Wikimedia’s arguments about the merits of Schulzrinne’s

opinions.

14 Judge Motz chides us for (as she describes it) rushing to

decide this issue in the face of the Supreme Court’s grant of

certiorari in Fazaga. But this case was briefed and argued

months before the Court decided to take Fazaga, and we have

given it all due deliberation. Moreover, our superior Court is

often informed by the views of the circuits. See, e.g., Hertz Corp.

v. Friend, 559 U.S. 77, 92 (“In an effort to find a single, more

uniform interpretation of the statutory phrase, we have reviewed

the Courts of Appeals’ divergent . . . interpretations.”). As we’ve

done in the past, we respectfully offer our perspective on this

“novel and difficult question” (Dissent at 56) before the Court

provides a definitive answer. See, e.g., Int’l Refugee Assistance

Project v. Trump, 883 F.3d 233 (4th Cir. 2018) (affirming the

district court’s grant of a preliminary injunction despite the

Supreme Court’s grant of a writ of certiorari on the same issues).

36a

The parties first debate the origin of the state

secrets privilege. Wikimedia calls it a common law

privilege, which Congress can abrogate by passing a

statute that “speak[s] directly to the question

addressed by” the privilege, even if the statute doesn’t

“affirmatively proscribe it.” United States v. Texas,

507 U.S. 529, 534 (1993). The government says the

privilege is “constitutionally grounded,” Appellee’s Br.

at 11, and can only be supplanted where “Congress

specifically has provided” for a statute to do so. Dep’t

of Navy v. Egan, 484 U.S. 518, 530 (1988).

We have indeed observed that the state secrets

privilege is an evidentiary rule “bas[ed] in the

common law of evidence.” El-Masri, 479 F.3d at 303–

04; see also Gen. Dynamics Corp. v. United States, 563

U.S. 478, 485 (2011) (“Reynolds . . . decided a purely

evidentiary dispute by applying evidentiary rules.”).

But we’ve also recognized that the privilege “performs

a function of constitutional significance[] because it

allows the executive branch to protect information

whose secrecy is necessary to its military and foreignaffairs responsibilities.” El-Masri, 479 F.3d at 303; see

also United States v. Nixon, 418 U.S. 683, 711 (1974)

(“[T]o the extent [an evidentiary privilege] relates to

the effective discharge of a President’s powers, it is

constitutionally based.”).

Fortunately, we need not decide today who has the

better argument. As we explain, even if we agree with

Wikimedia that the state secrets privilege is grounded

in the common law (which Congress may abrogate),

FISA doesn’t “speak directly” to the situation here.

Texas, 507 U.S. at 534.

37a

“We begin, as always in deciding questions of

statutory interpretation, with the text of the statute.”

Othi v. Holder, 734 F.3d 259, 265 (4th Cir. 2013). The

relevant subsection of FISA provides:

Whenever a court or other authority is

notified pursuant to subsection (c) or (d) of

this section, or whenever a motion is made

pursuant to subsection (e) of this section, or

whenever any motion or request is made by

an aggrieved person pursuant to any other

statute or rule of the United States or any

State before any court or other authority of

the United States or any State to discover or

obtain applications or orders or other

materials relating to electronic surveillance

or to discover, obtain, or suppress evidence or

information obtained or derived from

electronic surveillance under this chapter, the

United States district court or, where the

motion is made before another authority, the

United States district court in the same

district

the

authority,

shall,

notwithstanding any other law, if the

Attorney General files an affidavit under oath

that disclosure or an adversary hearing would

harm the national security of the United

States, review in camera and ex parte the

application, order, and such other materials

relating to the surveillance as may be

necessary to determine whether the

surveillance of the aggrieved person was

lawfully authorized and conducted. In making

this determination, the court may disclose to

the aggrieved person, under appropriate

security procedures and protective orders,

38a

portions of the application, order, or other

materials relating to the surveillance only

where such disclosure is necessary to make an

accurate determination of the legality of the

surveillance.

50 U.S.C. § 1806(f). FISA further defines an

“aggrieved person” as a “person who is the target of an

electronic surveillance or any other person whose

communications or activities were subject to

electronic surveillance.” Id. at § 1801(k).

The first lines of this subsection describe three

conditions that trigger the district court’s in camera

and ex parte review obligations. These are: (i) when

the federal or state government notifies the court that

it intends to use electronic surveillance information

against an aggrieved person, which it’s required to do

before introducing such evidence in a judicial

proceeding under § 1806(c) or (d); (ii) when an

aggrieved person makes a motion to suppress

electronic surveillance information used by the

government under § 1806(e); and (iii) when an

aggrieved person makes “any motion or request . . .

pursuant to any other statute or rule . . . to discover

or obtain applications or orders or other materials

relating to electronic surveillance or to discover,

obtain, or suppress evidence or information obtained

or derived from electronic surveillance.” Id. at §

1806(f).

Relying heavily on Fazaga, Wikimedia claims that

the third condition unambiguously encompasses the

circumstances at hand: Wikimedia is an aggrieved

person that made a motion before the district court

under Federal Rule of Civil Procedure 37(a) to compel

39a

discovery of “materials relating to electronic

surveillance.” Id. at § 1806(f). Wikimedia thus reads §

1806(f) as a free-floating right to obtain information

related to the government’s electronic surveillance

pursuant to any (and all) federal statutes or rules.

But “[t]he plainness or ambiguity of statutory

language is determined by reference to the language

itself, the specific context in which that language is

used, and the broader context of the statute as a

whole.” Healthkeepers, Inc. v. Richmond Ambulance

Auth., 642 F.3d 466, 471 (4th Cir. 2011) (quoting

Robinson v. Shell Oil Co., 519 U.S. 337, 340 (1997)).

“This includes employing various grammatical and

structural canons of statutory interpretation which

are helpful in guiding our reading of the text.” Id.

(citing United Sav. Ass’n. v. Timbers of Inwood Forest

Assocs., 484 U.S. 365, 371 (1988)).

Reading the third condition in context reveals that

Wikimedia’s gloss makes for a shiny but ill-fitting

shoe. Both parties agree that § 1806(f) may apply

regardless of who initiated the suit. But we agree with

the government that § 1806(f) describes procedures for

determining

the

admissibility

electronic

surveillance information only when the government

seeks to use such evidence in a particular

proceeding—whether civil or criminal. Thus, even

assuming that Wikimedia is an aggrieved person, 15

we conclude that it can’t use § 1806(f) to force the

government to introduce electronic surveillance

information into this case. To the extent our

15 Given our assumption, we don’t have to determine what a

litigant must prove to qualify as an aggrieved person and

whether Wikimedia has done so.

40a

reasoning, as laid out below, is inconsistent with

Fazaga, we decline to follow our sister circuit.

“[W]e rely on the principle of noscitur a sociis—a

word is known by the company it keeps—to avoid

ascribing to one word a meaning so broad that it is

inconsistent with its accompanying words, thus giving

unintended breadth to the Acts of Congress.” Yates v.

United States, 574 U.S. 528, 543 (2015) (cleaned up).

Both of the specific conditions in § 1806(f) (notice of

government intent to use surveillance information

and a motion to suppress) presume the government’s

introduction of surveillance evidence into the

proceedings. The subsequent general condition “is

therefore appropriately read to refer, not to any

[motion],” as Wikimedia asserts, “but specifically to

the subset” of motions contingent on the government’s

use of surveillance evidence. Id. at 544; see also id. at

543 (explaining that the meaning of a word in a list

may be limited by the other enumerated terms, “even

though the list began with the word ‘any’”).

Relatedly, where “general words follow specific

words in a statutory enumeration,” the ejusdem

generis canon counsels that “the general words are

construed to embrace only objects similar in nature to

those objects enumerated by the preceding specific

words.” Wash. State Dep’t of Soc. & Health Servs. v.

Guardianship Estate of Keffeler, 537 U.S. 371, 384

(2003); see also CSX Transp., Inc. v. Ala. Dept. of

Revenue, 562 U.S. 277, 295 (2011) (“We typically use

ejusdem generis to ensure that a general word will not

render specific words meaningless.”). Here, if §

1806(f)’s third condition requiring the district court to

act on “any motion or request” were as “all

encompassing” as Wikimedia alleges, it would render

the second condition superfluous. See Yates, 574 U.S.

41a

at 546. “Congress would have had no reason to refer

specifically to [motions to suppress]”—in fact, it’s

“hard to see why [Congress] would have needed to

include the examples at all.” Id. at 545–46.

It makes more sense to conclude that, by including

the two preceding conditions, Congress signaled its

intent to “cabin the contextual meaning” of the third

condition. Id. at 543. Like its predecessors, the third

condition thus applies only when an aggrieved person

makes a motion or request in response to the

government’s attempt to use surveillance evidence in

a proceeding.

This interpretation accords with the limitations

that Congress attached to the third condition on the

back end. Section 1806(f) specifies that the litigant’s

motion must be “to discover, obtain, or suppress.”

These are familiar “procedural motions pertaining to

the admissibility of evidence.” Fazaga, 965 F.3d at

1083 (Butamay, J. dissenting). The direct objects of

those actions are the “applications or orders or other

materials relating to electronic surveillance” or the

“evidence or information obtained or derived from

[such] surveillance.” 50 U.S.C. § 1806(f). And the

district court’s review is correspondingly restricted to

the “application, order, and such other materials

relating to the surveillance as may be necessary to

determine whether the surveillance . . . was lawfully

authorized and conducted.” Id.

Here, too, noscitur a sociis and ejusdem generis

color our understanding of “other material” and “such

other material” to mean those like a FISA application

or order—i.e., documents related to officially

approving and defining the scope of FISA surveillance

that can thus be used to determine the legality of the

42a

government’s surveillance operations. See also Such,

MERRIAM-WEBSTER,

https://www.merriamwebster.com/dictionary/such (last visited August 18,

2021) (defining “such” as “of the character, quality, or

extent previously indicated or implied”).

In short, the third condition in § 1806(f) is confined

to procedural requests related to a circumscribed body

evidence

(i.e.,

the

government’s

FISA

documentation and the resulting intelligence). This

corresponds with interpreting § 1806(f) as directed

towards determining the admissibility of the fruits of

the government’s surveillance—a question that arises

only when the government offers such evidence in a

case—and not as an unbounded invitation for litigants

to acquire any information they desire about the

government’s intelligence programs.

The remedy available to a successful movant

confirms our reading of this condition. As the very

next subsection provides:

If the United States district court pursuant to

subsection

(f)

determines

that

the

surveillance was not lawfully authorized or

conducted, it shall, in accordance with the

requirements of law, suppress the evidence

which was unlawfully obtained or derived

from electronic surveillance of the aggrieved

person or otherwise grant the motion of the

aggrieved person. If the court determines that

the surveillance was lawfully authorized and

conducted, it shall deny the motion of the

aggrieved person except to the extent that due

process requires discovery or disclosure.

43a

50 U.S.C. § 1806(g).

The paradigmatic remedy is thus the suppression

of evidence. It’s even the focus of the subsection’s title:

“Suppression of evidence; denial of motion.” Id.; see

also Fla. Dept. of Revenue v. Piccadilly Cafeterias,

Inc., 554 U.S. 33, 47 (2008) (“[A] subchapter heading

cannot substitute for the operative text of the statute.

Nonetheless, statutory titles and section headings are

tools available for the resolution of a doubt about the

meaning of a statute.” (cleaned up)). 16 And a litigant

would seek such a remedy only in response to the

government’s introduction of surveillance evidence

into the case.

By contrast, consider the mismatch between the

remedy described in § 1806(g) and the remedy that

Wikimedia seeks. Rather than request the

suppression of evidence, Wikimedia wants the district

court to review the evidence requested by the motion

to compel to decide both standing and the merits of its

unlawful surveillance claims. But that approach

would contort the §1806(f) and (g) procedures beyond

recognition. The statutory text doesn’t permit the

district court to rule on anything other than the

motion at hand or consider evidence beyond the FISA

application and related materials, let alone conduct

an entire trial in camera and grant final judgment on

the merits of the underlying claim. 17

16 The titles of § 1806 as a whole, “Use of information,” and §

1806(f), “In camera and ex parte review by district court,” are less

illuminating but remain consistent with the notion that it’s the

government’s use of information that matters.

Wikimedia argues that this emphasis on motions to

suppress is misguided because § 1806(f) expressly includes more

44a

We note that every other subsection under § 1806

speaks to the government’s use of electronic

surveillance evidence. Section 1806(a) provides that

such evidence “may be used and disclosed by Federal

officers and employees” in compliance with

minimization procedures; (b) says that such evidence

“may only be used in a criminal proceeding with the

advance authorization of the Attorney General”; (c)

and (d) mandate that federal and state governments

give notice before using such information against an

aggrieved person; (e) permits an aggrieved person to

file a motion to suppress such evidence when used

against him; (i) requires that the government destroy

information unintentionally acquired through

electronic surveillance; (j) instructs a court about

notifying an aggrieved person when the government

conducts emergency surveillance without preauthorization; and (k) allows federal officers who

conduct electronic surveillance to coordinate with

federal or state law enforcement officers. 18 We think

it unlikely that Congress stashed away an expansive

right for litigants within a statute directed entirely

toward the government’s use of information. See

Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 468

(2001) (“Congress . . . does not, one might say, hide

elephants in mouseholes.”).

Still, Wikimedia resists this interpretation,

contending that the government’s reading effectively

than that. Even accepting that as true, we are confident that it

doesn’t contemplate what Wikimedia seeks in this litigation.

Section 1806(h) is the only subsection that doesn’t

expressly relate to the government’s use of information, but it

merely provides that a district court’s decisions under subsection

(g) are final and binding upon all other federal courts.

45a

means that a plaintiff can only rely on § 1806(f) after

the government has given notice that it’s using

electronic surveillance information per § 1806(c) or

(d), 19 which renders the two other conditions for

obtaining in camera and ex parte review superfluous.

That might be the case if the government were always

scrupulous in providing such notice. But even the

government admits that there has been some

“dispute” about its withholding of notice in the past,

though it claims to have “redoubled its efforts” since

the Solicitor General’s 2013 confession of error on this

front. Oral Argument at 35:07–35:56.

It’s therefore reasonable for Congress to have

crafted additional paths for ascertaining the legality

of electronic surveillance evidence that the

government intends to marshal against a litigant who

can show that it is an “aggrieved person,” even when

the government has violated its duty to provide notice

of such use. Cf. United States v. Belfield, 692 F.2d 141,

146 (D.C. Cir. 1982) (explaining that a litigant

“claim[ing] that he has been the victim of an illegal

surveillance [operation] and seek[ing] discovery of the

[surveillance records] to ensure that no fruits thereof

are being used against him” can trigger the § 1806(f)

procedures even though the government “has

purported not to be offering any [such] evidence”).

Additionally, Wikimedia asserts that the phrase

“notwithstanding any other law” is an indication that

FISA displaces the state secrets privilege. But that

clause applies only when the plaintiff has fulfilled one

19 As mentioned above, § 1806(c) and (d) require federal and

state governments, respectively, to give notice to the court or to

the aggrieved person when they intend to use surveillance

evidence against such a person in a judicial proceeding.

46a

of the three pre-requisite conditions for triggering the

court’s in camera and ex parte review, and the

Attorney General has filed the necessary affidavit.

Only then “shall” the court apply the §1806(f) in

camera procedures, “notwithstanding any other law”

that would require some other, public resolution of the

litigant’s motion challenging the government’s use of

electronic surveillance information. See S. Rep. No.

95-701, at 63 (“Although a number of different

procedures might be used to attack the legality of the

surveillance, it is this procedure ‘notwithstanding any

other law’ that must be used to resolve the question. .

. . This is necessary to prevent the carefully drawn

procedures in subsection [(f)] from being bypassed by

the inventive litigant using a new statute, rule or

judicial construction.”). 20

And although § 1806(f) and the state secrets

privilege are triggered by an affidavit from the

government, it doesn’t follow that FISA speaks

directly to the state secrets privilege, as Wikimedia

claims. Tellingly, these procedures contemplate

different affiants. Because the privilege is a shield to

protect state secrets from disclosure, the head of the

department controlling the information must assert

it. By contrast, FISA applies when the government is

attempting to offer electronic surveillance evidence in

a case. In such an instance, responsibility for invoking

§ 1806(f) falls to the one who wields the sword: the

Attorney General (or his delegees, under 50 U.S.C. §

20 In the draft of the statute discussed by this report, what is

now subsection (f) was located under subsection (e). See S. Rep.

No. 95-701, at 88. Despite the different lettering, the substance

of the provision was largely the same. We have edited the quote

to correspond with the current organization of § 1806’s

provisions.

47a

1801(g)). The triggering mechanisms for each

procedure thus strengthen the inference that FISA

wasn’t intended to displace the state secrets privilege.

Wikimedia further contends that limiting the

applicability of § 1806(f) and (g) to when the

government offers electronic surveillance evidence in

a case is inconsistent with FISA as a whole. In

particular, Wikimedia complains that if § 1806(f)

doesn’t displace the state secrets privilege, the

government can invoke the privilege “in every FISA

suit brought by a civil plaintiff.” Reply Br. at 5. This

would, in turn, give the government “nearly exclusive

control over challenges to FISA surveillance” and

“profoundly undermine the civil remedies that

Congress enacted for surveillance abuses[] and the

very purpose of FISA itself,” which Wikimedia asserts

is “to ensure judicial review of executive branch

surveillance.” Appellant’s Br. at 51–52.

We are not convinced. The government knows that

it carries the burden “to satisfy the reviewing court

that the Reynolds reasonable-danger standard is

met,” Abilt v. CIA, 848 F.3d 305, 312 (4th Cir. 2017)

(cleaned up), and that the judiciary is “firmly in

control of deciding whether an executive assertion of

the state secrets privilege is valid,” El-Masri, 479 F.3d

at 304–05. Indeed, the court stands as a gatekeeper to

the privilege, and “[w]e take very seriously our

obligation to review the [government’s claims] with a

very careful, indeed a skeptical, eye,” Abilt, 848 F.3d

at 312 (quoting Al–Haramain Islamic Found., Inc. v.

Bush, 507 F.3d 1190, 1203 (9th Cir. 2007)), so that

“the state secrets privilege is asserted no more

frequently and sweepingly than necessary,” id.

48a

(quoting Ellsberg v. Mitchell, 709 F.2d 51, 58 (D.C.

Cir. 1983)). There have thus been FISA cases where

the government hasn’t invoked the privilege, see, e.g.,

Am. Civil Liberties Union v. Clapper, 785 F.3d 787,

801–03 (2d Cir. 2015), or has invoked the privilege

narrowly, see, e.g., Fazaga, 965 F.3d at 1042 (“Here,

although the Government has claimed the Reynolds

privilege over certain state secrets, it has not sought

dismissal of the Fourth Amendment and FISA claims

based on its invocation of the privilege.”).

Nor do we see any actual contradictions between

FISA and the Reynolds privilege. Congress provided

for judicial review of executive branch surveillance,

but it did so to “strike[] a fair and just balance between

protection of national security and protection of

personal liberties.” S. Rep. No. 95-604, pt. 1, at 7

(1978). The government’s reading of § 1806(f) fits that

schema exactly. In that provision, Congress permits

the government to use electronic surveillance

evidence in court against a litigant while withholding

materials related to that surveillance from that

individual in the interests of national security. But in

the same breath, Congress also allows an aggrieved

person to challenge the government’s use of such

evidence and have a court evaluate the lawfulness of

the government’s actions.

Far from giving the government exclusive control

over challenges to surveillance, we think this reading

of § 1806(f) acknowledges the court’s role in

preserving the compromise Congress made between

individual rights and national security. See Belfield,

692 F.2d at 149 (“If anything, the legality inquiry

mandated by FISA is easier for a court to perform ex

parte than the pre-FISA inquiry into the legality of

warrantless electronic surveillance . . .”). For instance,

49a

when “the Court believes that disclosure is necessary

to make an accurate determination of legality, but the

Government argues that to do so . . . would damage

the national security,” § 1806(f) says that “the

Government must choose— either disclose the

material or forgo the use of the surveillance-based

evidence.” See S. Rep. No. 95-701, at 65.

Additionally, judicial review occurs at another

point in the FISA process. “Congress created a

comprehensive scheme in which the [FISC] evaluates

the Government’s certifications, targeting procedures,

and minimization procedures— including assessing

whether the targeting and minimization procedures

comport with the Fourth Amendment,” Clapper, 568

U.S. at 421, which we described more fully in our prior

opinion, see Wikimedia Found., 857 F.3d at 200–01.

“Any dissatisfaction that [Wikimedia] may have about

the [FISC]’s rulings—or the congressional delineation

of that court’s role—is irrelevant” to our analysis.

Clapper, 568 U.S. at 421.

In sum, the government’s reading of § 1806 is

entirely consistent with ensuring judicial review of

executive branch surveillance. That’s not surprising

considering the history of courts uniformly using in

camera procedures to determine the legality of

foreign- intelligence surveillance even before FISA’s

enactment. See Belfield, 692 F.2d at 149 & n.38

(collecting cases). As the government observes, “[t]hat

such [in camera] procedures comfortably coexisted

with the [state secrets] privilege before FISA

underscores that codification of in camera procedures

for certain purposes,” without more, doesn’t suggest

that Congress intended to displace the privilege.

Appellee’s Br. at 35; see also H.R. Rep. No. 95-1283

(1978) (“[O]nce the surveillance is determined to be

50a

unlawful, the intent of [§ 1806] is to leave to otherwise

existing law the resolution of what, if anything, is to

be disclosed.”).

The only “inconsistency” between FISA and the

state secrets privilege Wikimedia identifies is that

Congress provided civil remedies for violations of

FISA that a plaintiff may have to forego when the

government invokes the Reynolds privilege. These

include 50 U.S.C. § 1810, whereby a plaintiff may

recover damages from a person who is criminally

prosecuted under 50 U.S.C. § 1809 for intentionally

engaging in, disclosing, or using electronic

surveillance in violation of FISA; and 18 U.S.C. §

2712, which permits a plaintiff to recover damages

from the United States for a willful violation of FISA.

But this problem isn’t unique to FISA. Every state

secrets case presents the possibility that a plaintiff

will be denied—in the interests of national security—

a remedy available by law. See El-Masri, 479 F.3d at

313 (“[T]he successful interposition of the state secrets

privilege imposes a heavy burden on the party against

whom the privilege is asserted . . . not through any

fault of his own, but because his personal interest in

pursuing his civil claim is subordinated to the

collective interest in national security.”); Fitzgerald v.

Penthouse Int’l, Ltd., 776 F.2d 1236, 1238 n.3 (“When

the state secrets privilege is validly asserted, the

result is unfairness to individual litigants—through

the loss of important evidence or dismissal of a case—

in order to protect a greater public value.”); Sterling v.

Tenet, 416 F.3d 338, 348 (4th Cir. 2005) (“[I]n limited

circumstances like these, the fundamental principle of

access to courts must bow to the fact that a nation

without sound intelligence is a nation at risk.”).

51a

Accordingly, we conclude that § 1806(f) doesn’t

displace the state secrets privilege, even in actions

pertaining to government-run electronic surveillance.

Because FISA’s discovery procedures don’t govern

here, we turn to whether the district court properly

applied the state secrets privilege. We hold that the

privilege indeed requires dismissal of this case.

When a state secrets question arises, a court

applies a three-part analysis. First, “the court must

ascertain that the procedural requirements for

invoking the state secrets privilege have been

satisfied”—i.e., that the government properly made a

formal claim of privilege. El-Masri, 479 F.3d at 304.

Wikimedia doesn’t dispute that the government

satisfied this condition.

Second, “the court must decide whether the

information sought to be protected qualifies as

privileged” because it is a state secret. Id. That is, it

must determine, “from all the circumstances of the

case,” whether “there is a reasonable danger that

compulsion of the evidence will expose military

matters which, in the interest of national security,

should not be divulged.” Reynolds, 345 U.S. at 10.

This inquiry “pits the judiciary’s search for truth

against the Executive’s duty to maintain the nation’s

security.” El-Masri, 479 F.3d at 305. Accordingly,

“[t]he degree to which such a reviewing court should

probe depends in part on the importance of the

assertedly privileged information to the position of the

party seeking it”: “where there is a strong showing of

necessity, the claim of privilege should not be lightly

52a

accepted,” but “even the most compelling necessity

cannot overcome the claim of privilege if the court is

ultimately satisfied that [state] secrets are at stake.”

Id.

The district court found that “the entities subject

to Upstream surveillance activity and the operational

details of the Upstream collection process” were state

secrets because their disclosure “would (i) undermine

ongoing intelligence operations, (ii) deprive the NSA

of existing intelligence operations, and significantly,

(iii) provide foreign adversaries with the tools

necessary both to evade U.S. intelligence operations

and to conduct their own operations against the

United States and its allies.” Wikimedia Found., 427

F. Supp. 3d at 613. The court thus sustained the

government’s claim of privilege for the seven

categories of information identified by the NSA.

Wikimedia doesn’t meaningfully dispute the

court’s finding on this prong either. Instead, it

quibbles that to the extent the seven categories cover

material that the government has already disclosed,

the district court’s ruling is overly broad. But we don’t

read the court’s decision to make privileged what’s

already public. The court instead concluded that

because of the state secrets privilege, Wikimedia

couldn’t compel the government to produce, or

otherwise continue to pursue litigation that would

risk the disclosure of, additional information related

to those categories. See Wikimedia Found., 427 F.

Supp. 3d at 611.

And based on the totality of the circumstances,

including the Coats and Barnes affidavits, we agree

that “there is a reasonable danger” to national

53a

security should these facts be disclosed. El-Masri, 479

F.3d at 305; see also id. (“Frequently, the explanation

of the department head who has lodged the formal

privilege claim . . . is sufficient to carry the Executive’s

burden.”).

This leads us to the third step, which is to resolve

“how the matter should proceed in light of the

successful privilege claim.” El-Masri, 479 F.3d at 304.

Once a court determines that certain facts are state

secrets, they are “absolutely protected from

disclosure.” Id. at 306. “[N]o attempt is made to

balance the need for secrecy of the privileged

information against a party’s need for the

information’s disclosure.” Id.

As a result, “[i]f a proceeding involving state

secrets can be fairly litigated without resort to the

privileged information, it may continue.” Id. But if

“any attempt to proceed will threaten disclosure of the

privileged matters, dismissal is the proper remedy.”

Id. (cleaned up). The latter situations include where:

(1) “the plaintiff cannot prove the prima facie

elements of his or her claim without privileged

evidence”; (2) “even if the plaintiff can prove a prima

facie case without resort to privileged information, . .

. the defendants could not properly defend themselves

without using privileged evidence”; and (3) “further

litigation would present an unjustifiable risk of

disclosure.” Abilt, 848 F.3d at 313–14. 21

21 Wikimedia asserts (without further explanation) that the

third basis for invoking the state secrets privilege “wrongly

collapses the Reynolds privilege and the Totten [v. United States,

92 U.S. 105 (1876)] bar.” Appellant’s Br. at 58 n.19 (citing Gen.

Dynamics, 563 U.S. at 485). “Totten has come to primarily

represent . . . a categorical bar on actions to enforce secret

54a

Here, the district court determined that both the

second and third situations apply such that “dismissal

is the appropriate, and only available, course of

action.” Wikimedia Found., 427 F. Supp. 3d at 611.

Wikimedia now argues that because it established a

prima facie case for standing using public evidence,

the court should have reviewed the purportedly

privileged material in camera to determine the

validity—or at least the existence—of the

government’s hypothetical defense before ordering the

case dismissed.

We agree with the district court that in camera

review in this instance would fly in the face of the

state secrets privilege as espoused by “both Supreme

Court precedent and our own cases.” Sterling, 416

F.3d at 345. A district court may consider any

evidence it deems necessary at step two of the

Reynolds inquiry—i.e., when determining whether

the information at issue comprises state secrets. See

id. (“There may . . . be cases where the necessity for

This text is long and has been trimmed here. Open the source document for the complete record.

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.