PUBLISHED
UNITED STATES COURT OF APPEALS
FOR THE FOURTH CIRCUIT

No. 20-1495

LEADERS OF A BEAUTIFUL STRUGGLE; ERRICKA BRIDGEFORD; KEVIN
JAMES,
Plaintiffs – Appellants,
v.
BALTIMORE POLICE DEPARTMENT; MICHAEL S. HARRISON, in his official
capacity as Baltimore Police Commissioner,
Defendants - Appellees.

Appeal from the United States District Court for the District of Maryland, at Baltimore.
Richard D. Bennett, District Judge. (1:20-cv-00929-RDB)

Argued: September 10, 2020

Decided: November 5, 2020

Before GREGORY, Chief Judge, and WILKINSON and NIEMEYER, Circuit Judges.

Affirmed by published opinion. Judge Wilkinson wrote the opinion, in which Judge
Niemeyer joined. Chief Judge Gregory wrote a dissenting opinion.

ARGUED:
Brett Max Kaufman, AMERICAN CIVIL LIBERTIES UNION
FOUNDATION, New York, New York, for Appellants.
Rachel Simmonsen,
BALTIMORE CITY LAW DEPARTMENT, Baltimore, Maryland, for Appellees. ON
BRIEF: David R. Rocah, AMERICAN CIVIL LIBERTIES UNION FOUNDATION OF
MARYLAND, Baltimore, Maryland; Ashley Gorski, Alexia Ramirez, Nathan Freed
Wessler, Ben Wizner, AMERICAN CIVIL LIBERTIES UNION FOUNDATION, New
York, New York, for Appellants.

 WILKINSON, Circuit Judge:
Plaintiffs appeal the district court’s denial of their request for a preliminary
injunction against Baltimore’s aerial surveillance program. For the reasons set forth herein,
we conclude that the district court did not abuse its discretion in denying the preliminary
injunction.
I.
Baltimore sadly has experienced a serious recent rise in homicides. For each of the
past five years, Baltimore has been victimized by at least three hundred murders. In 2017,
Baltimore experienced a higher absolute number of murders than New York City, a city
with fourteen times Baltimore’s population. See Alec MacGillis, The Tragedy of Baltimore,
N.Y. Times Magazine, Mar. 17, 2019, at 32. Moreover, the Baltimore Police Department
(BPD) has struggled to respond effectively to this increase in murders. In 2019, it cleared
just 32.1% of homicide investigations, its lowest rate in several decades. Jessica Anderson,
Baltimore Ending the Year with 32% Homicide Clearance Rate, One of the Lowest in Three
Decades, Balt. Sun, Dec. 30, 2019.
As Judge Bennett noted during the preliminary injunction hearing, “even a
pandemic cannot slow the pace of killings” in Baltimore. Transcript of Preliminary Hearing
at 74, Leaders of a Beautiful Struggle v. Balt. Police Dep’t, No. RDB-20-0929, 2020 WL
1975380 (D. Md. Apr. 24, 2020). As of April 21, the date of the district court’s hearing,
Baltimore was on pace for a higher number of murders this year than in 2019. Leaders of
a Beautiful Struggle v. Balt. Police Dep’t, No. RDB-20-0929, 2020 WL 1975380, at * 15
(D. Md. Apr. 24, 2020).
2

 One step taken by the BPD to strengthen its hand against violent crime is the Aerial
Investigative Research program (AIR). It is a carefully limited program of aerial
observations of public movements presented as dots, and it is important at the outset to say
all the things the program does not do. It does not search a person’s home, car, personal
information or effects. It does not photograph a person’s features. The program has been
progressively circumscribed to meet the thoughtful objections of civil libertarians, though
not sufficiently in plaintiffs’ view.
To implement and test this program, the BPD has partnered with a private company,
Persistent Surveillance Systems (PSS). The program operates by flying three small planes
over Baltimore during daytime hours, weather permitting. The planes are equipped with
cameras that cover about ninety percent of the city at any given time. The cameras employ
a resolution that reduces each individual on the ground to a pixelated dot, thus making the
cameras unable to capture identifying characteristics of people or automobiles.
The resulting photographs are transmitted to a control room, staffed by fifteen to
twenty-five analysts employed by PSS along with BPD officers. The control room can
access the photographs only when specific violent crimes—shootings, robberies, and
carjackings—are reported in a particular location. Upon receiving a notification, analysts
can “tag” the dots photographed around the crime scene and track those dots’ public
movements in the hours leading up to and following the crime. The process takes about
eighteen hours and provides the BPD with a report that includes the location and timing of
the crime, the observable actions at the crime scene, the tracks of people and vehicles to
and from the crime scene, and the locations the individuals at the crime scene visited before
3

 and after the crime. Using that data, the police can employ existing surveillance tools, such
as on-the-ground surveillance cameras and license-plate readers, to identify witnesses and
suspects. Within seventy-two hours, analysts can give the police a more detailed report
about those present at the crime scene that potentially includes identifying information.
The partnership between PSS and the BPD began in 2016, when AIR was used to
obtain about three hundred hours of surveillance. The BPD did not disclose the privately
funded program’s existence to the public, Baltimore’s elected officials, or even the city
solicitor. Then-Police Commissioner Kevin Davis later stated that he wanted to see if the
program worked before bringing it to the City Council. After a news article exposed the
program’s existence, it was temporarily shut down.
Under a new police commissioner, defendant Michael Harrison, the BPD moved to
reactivate AIR and introduced safeguards intended to guard against potential abuses. In the
name of transparency, the BPD has made public its contract with PSS and hosted public
fora via social media networks like Facebook to explain and answer questions about AIR.
Further, the BPD rallied substantial political support for the project, including from
Governor Larry Hogan, dozens of crime victims, community groups, business advocacy
groups, and the United Baptist Ministry Convention. The BPD has also adopted several
limitations on the collection, use, and retention of photographs obtained through AIR
surveillance in its contract with PSS. See J.A. 69–73. Those limitations are as follows:
• AIR’s planes fly only the daytime hours, weather permitting, and never at
night.

4

 • AIR uses limited resolution cameras that identify individuals only as
pixelated dots in a photograph. Analysts examining these photographs are
not able to identify an individual’s race, gender, or clothing.
• If a dot is seen entering a building in a photograph, analysts cannot know if
the same person is leaving the building when they see a dot leave the building
without the use of other surveillance tools.
• The cameras do not utilize zoom, infrared, or telephoto technologies.
• Analysts cannot access photographs until they receive a notification related
to the investigation of a specific murder, non-fatal shooting, armed robbery,
or carjacking.
• There will be no live tracking of individuals. Analysts can only use AIR’s
photographs to look at past movements.
• If an arrest is made using the AIR technology, the photos related to the arrest
will be given to the prosecutor and defense counsel. Otherwise, all
photographs collected by AIR will be deleted after forty-five days.
After the district court denied the plaintiffs’ request for a preliminary injunction,
AIR became operational. It is currently near the completion of a six-month testing period,
at the end of which its continued use will be assessed. As part of this assessment, the BPD
has invited several independent research partners to study the program. The RAND
Corporation will study how effective the program is at solving crimes. The Policing Project
at the New York University School of Law will assess possible civil rights and civil
liberties concerns. The University of Baltimore will study how residents of Baltimore
perceive the program and whether, if at all, the program affects views of police legitimacy.
II.
On April 9, 2020, plaintiffs sued the BPD and Commissioner Harrison under 42
U.S.C. § 1983 challenging the constitutionality of the AIR program under the First and
5

 Fourth Amendments. They alleged that the program would infringe their reasonable
expectations of privacy because the police were likely to generate activity reports about
them, that it would hinder their work in the community by making people afraid to talk to
them on the streets for fear of being surveilled, and that it would chill their First
Amendment rights by making them more hesitant to associate with people for fear of being
observed. They sought a preliminary injunction.
After a hearing, the district court judge denied plaintiffs’ request for a preliminary
injunction. After finding that plaintiffs had standing, the court concluded that a preliminary
injunction was inappropriate because plaintiffs were not likely to succeed on the merits of
their complaint. Leaders of a Beautiful Struggle, 2020 WL 1975380 at *8–13. The court
discussed the various technical limitations of the AIR program, including that the planes
can only fly twelve hours and only during the day (weather permitting) and that individuals
would only be represented as pixelated dots to the persons looking at the captured images.
Id. at *4. The upshot of these limitations, Judge Bennett explained, was that individuals’
movements could not be tracked over the course of multiple days. Id. at *12. He reasoned
that the Supreme Court and the Fourth Circuit had upheld more intrusive forms of aerial
surveillance, and that the Supreme Court’s decision in Carpenter v. United States, 138 S.
Ct. 2206 (2018), did not abrogate those precedents. Leaders of a Beautiful Struggle, 2020
WL 1975380 at *9–12.
III.
As a preliminary matter, we reject the defendants’ argument that plaintiffs lack
standing to bring this lawsuit. Article III of the Constitution limits the judicial power of the
6

 United States to “Cases” and “Controversies.” U.S. Const. art. III, § 2. To implement this
limitation on judicial power, the Supreme Court has explained that a litigant must show an
injury in fact, that the injury is caused by the defendant’s unlawful conduct, and that the
injury is redressable by a favorable judicial ruling. Steel Co. v. Citizens for a Better Env’t,
523 U.S. 83, 103 (1998).
Defendants argue that plaintiffs’ alleged injuries are too speculative to establish
standing. They reason that the potential injury lies at the end of an attenuated chain of
possibilities where the BPD successfully identifies who plaintiffs are. But the injury
plaintiffs complain of is not being identified by the BPD, but merely that they are being
photographed. And that is almost certain to occur, considering that AIR’s cameras can
capture around ninety percent of Baltimore in a given moment.
In Wikimedia Foundation v. National Security Agency, 857 F.3d 193 (4th Cir. 2017),
this court found a concrete injury in a very similar case. There the plaintiffs alleged that
the government was unlawfully intercepting and copying their Internet communications.
Id. at 209–10. The government argued the plaintiffs lacked standing because they did not
allege the NSA’s collection involved human review, making it speculative the plaintiffs
would be identified as their communications were collected. But this court found
standing—the fact that the NSA collected and copied the communications was sufficient.
Id. at 209–11. So too here. It is enough to confer standing that plaintiffs are likely to be
photographed.

7

 IV.
By asking us to reverse the district court’s denial of a preliminary injunction,
plaintiffs face a doubly heavy burden on appeal. First, we review a district court’s denial
of a preliminary injunction for abuse of discretion. See Gonzales v. O Centro Espirita
Beneficente Uniao do Vegetal, 546 U.S. 418, 428 (2006). This deference recognizes the
traditional power district courts have had to craft remedies appropriate “to the necessities
of the particular case,” Weinberger v. Romero-Barcelo, 456 U.S. 305, 312 (1982), and it
reflects a commitment by appellate courts not to second-guess a district court’s remedial
decision absent special circumstances.
Second, plaintiffs seek an extraordinary remedy that should “be granted only
sparingly and in limited circumstances.” MicroStrategy Inc. v. Motorola, Inc., 245 F.3d
335, 339 (4th Cir. 2001) (internal quotation marks omitted). As the Supreme Court
explained in Winter v. Natural Resources Defense Council, Inc., 555 U.S. 7 (2008), “[a]
preliminary injunction is an extraordinary remedy never awarded as of right.” Id. at 24.
This principle reflects the reality that courts are more likely to make accurate decisions
after the development of a complete factual record during the litigation. DOUGLAS
LAYCOCK & RICHARD L. HASEN, MODERN AMERICAN REMEDIES 453 (5th ed. 2019).
To satisfy this standard, plaintiffs must make a “clear showing” that they satisfy
four independent requirements. Winter, 555 U.S. at 22; see also Real Truth About Obama,
Inc. v. Fed. Election Comm’n, 575 F.3d 342, 346 (4th Cir. 2009) (clarifying that all four
requirements must be independently satisfied by the party seeking a preliminary
injunction). Plaintiffs must demonstrate that: (1) they are likely to succeed on the merits of
8

 their legal claims; (2) they are likely to suffer irreparable harm in the absence of
preliminary relief; (3) the balance of equities favors the grant of preliminary relief; and (4)
that “an injunction is in the public interest.” Winter, 555 U.S. at 20. These requirements
make clear that preliminary injunctions should not be casually awarded.
We shall first address plaintiffs’ likelihood of success on the merits.
A.
The Fourth Amendment states that “[t]he 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.” U.S. Const. amend. IV. In evaluating whether a search is reasonable, the
Supreme Court has established a variety of doctrinal tests to be used depending on the
specific context of the search. The reasonable expectation of privacy test is the Court’s
most prominent standard, and we will evaluate the AIR program under it.
The Court has also tailored its standard to programs of searches designed to solve
important public safety problems. Mindful that, “[a]s the text of the Fourth Amendment
indicates, the ultimate measure of the constitutionality of a government search is
‘reasonableness,’” Maryland v. King, 569 U.S. 435, 447 (2013) (quoting Veronica School
Dist. 47J v. Acton, 515 U.S. 646, 652 (1995)), courts are sometimes required to balance
the degree of invasion of Fourth Amendment rights against the public safety interests
furthered when assessing the reasonableness of a surveillance program. The Supreme Court
has held that this standard applies to assess drunk driving checkpoints, see Michigan Dep’t
9

 of State Police v. Sitz, 496 U.S. 444 (1990), suspicionless searches at the border, see United
States v. Flores-Montano, 541 U.S. 149 (2004), and programmatic searches in prisons. See
Florence v. Bd. of Chosen Freeholders, 566 U.S. 318 (2012). Therefore, we will also
evaluate the AIR program under the balancing test used for programmatic searches.
We conclude initially that the AIR program does not violate a reasonable
expectation of privacy. This test has both a subjective and an objective component. The
court must first ask whether an individual actually expects that something will be kept
private from surveillance and then whether society is prepared to recognize that expectation
as reasonable. Carpenter, 138 S. Ct. at 2213. “Although no single rubric definitively
resolves which expectations of privacy are entitled to protection, the analysis is informed
by historical understandings ‘of what was deemed an unreasonable search and seizure
when [the Fourth Amendment] was adopted.’” Id. at 2213–14 (quoting Carroll v. United
States, 267 U.S. 132, 149 (1925)). The Court recognized two “guideposts” to assist the
inquiry. First, a court should ask whether the government action arbitrarily invades the
privacies of life. Id. at 2214. Second, courts should be wary of “a too permeating police
surveillance.” Id. (quoting United States v. Di Re, 332 U.S. 581, 595 (1948)).
A cardinal rule that emerges from the Supreme Court’s caselaw is that an individual
has a limited expectation of privacy in his or her public movements. “What a person
knowingly exposes to the public . . . is not a subject of Fourth Amendment protection.”
Katz v. United States, 389 U.S. 347, 351 (1967). Thus, “[a] person travelling in an
automobile on public thoroughfares has no reasonable expectation of privacy in his
movements from one place to another.” United States v. Knotts, 460 U.S. 276, 281 (1983).
10

 In light of that rule, the Supreme Court has found no Fourth Amendment violation when
the police used a tracking device placed inside a container to track an automobile on public
streets. Id. at 285.
In United States v. Jones, 565 U.S. 400 (2012), five Justices qualified this rule
through separate opinions by Justices Alito and Sotomayor. These two concurring opinions
reasoned that long-term surveillance using GPS tracking violated a reasonable expectation
of privacy. Id. at 414–15 (Sotomayor, J., concurring); id. at 430–31 (Alito, J., concurring
in the judgment) (distinguishing between a long-term surveillance using GPS for twentyeight days, which he thought was impermissible, and a shorter-term surveillance of public
movements).
The lesson from these cases is that short-term surveillance of an individual’s public
movements is less likely to violate a reasonable expectation of privacy. And under that
rule, the AIR program passes muster. As Judge Bennett explained, the built-in limitations
of the AIR program mean that it only enables the short-term tracking of public movements.
First, the AIR program’s cameras are only able to track outdoor movements. They cannot
track an individual who enters a building, and analysts cannot tell if the person leaving the
building is the same person who entered it. Second, AIR’s surveillance planes only fly
during twelve daylight hours. Because they do not fly at night, AIR surveillance cannot be
used to track individuals from day-to-day.
In response, plaintiffs object that the police may be able to use preexisting
surveillance tools, like security cameras and license plate readers, in conjunction with AIR
photographs to track individuals from day to day. But plaintiffs do not challenge these
11

 existing tools—only the AIR program in particular. And for good reason. The Supreme
Court specifically stated that traditional surveillance tools, specifically security cameras,
remain lawful in light of Carpenter, 138 S. Ct. at 2220, and we are not at liberty to revisit
that conclusion.
Further, the Supreme Court has already held that various types of aerial surveillance
do not invade a reasonable expectation of privacy. In California v. Ciraolo, 476 U.S. 207
(1986), the government flying a plane a mere one thousand feet over an individual’s home
to photograph items within a fenced-in backyard did not violate a reasonable expectation
of privacy. Id. at 209. The Court held that this intrusive surveillance was permissible as
long as the surveillance plane remained “within public navigable airspace” and the
surveillance was done “in a physically nonintrusive manner.” Id. at 213. Both of those
requirements are satisfied by AIR. Indeed, the AIR cameras cannot photograph an
individual item within a backyard—the program is decidedly less intrusive than the
surveillance in Ciraolo.
Similarly, in Dow Chemical Co. v. United States, 476 U.S. 227 (1986), the Supreme
Court found no invasion of a reasonable expectation of privacy when the government
repeatedly flew a plane over a factory and photographed items as small as one half inch in
diameter. Id. at 239. In this case, AIR does not use cameras capable of photographing
objects that small. Not even close. Individuals are just dots, and analysts cannot identify a
person’s race or sex from an AIR photograph. And like the surveillance in that case, AIR
captures no “intimate details” of the sort that might implicate Fourth Amendment
protections. Id. at 238.
12

 Finally, in Florida v. Riley, 488 U.S. 445 (1989), the Court again upheld
surveillance more intrusive than that in AIR. There, government agents circled four
hundred feet above a home in a helicopter to look into a greenhouse partially within the
home’s curtilage. Id. at 450. AIR involves no surveillance of an individual’s home or
curtilage, and its aerial planes do not harass individual homeowners by hovering closely
above their homes.
In light of these precedents, we cannot hold that AIR violates a reasonable
expectation of privacy. AIR is merely a tool used to track short-term movements in public,
where the expectation of privacy is lessened. Such an activity is lawful in light of Knotts
and Jones. And the specific tool which the BPD will use for the surveillance, aerial
photography, has been sanctioned by the Supreme Court in several cases. See also Brandon
Nagy, Why They Can Watch You: Assessing the Constitutionality of Warrantless
Unmanned Aerial Surveillance by Law Enforcement, 29 BERKELEY. TECH. L.J. 135, 155
(2014) (“[T]he Supreme Court's guidance, as derived from Riley, Dow, and Ciraolo,
appears to be that aerial surveillance, whether by naked eye or assisted by some technology,
conducted from lawful airspace that is traversed by aircraft with sufficient regularity, and
that does not interfere with the victim's use of her property, is not a Fourth Amendment
search and does not require a warrant.”). Unlike the aerial surveillance previously upheld,
AIR photographs cannot even be used to identify specific items within backyards or a
person’s identifying characteristics.
The Supreme Court’s decision in Carpenter does not require a different result. In
Carpenter, the Supreme Court considered whether targeting an individual and tracking his
13

 specific movements over a six-day period invaded a reasonable expectation of privacy. 138
S. Ct. at 2217. The technology at issue in Carpenter was cell-site location information
(CSLI), which was obtained because cell phones report their location to wireless towers
several times every minute. Id. at 2211–12. In Carpenter’s case, the government had, on
average, 101 location data points for each of the seven days it obtained CSLI data. Because
“a phone goes wherever its owner goes,” CSLI provides a “comprehensive record of the
person’s movements.” Id. at 2217. In summary, the Court held, CSLI “is detailed,
encyclopedic, and effortlessly compiled.” Id. at 2216. But the Court also emphasized the
narrowness of its holding, insisting it did not “call into question conventional surveillance
techniques and tools, such as security cameras.” Id. at 2220.
The district court recognized the important technological differences between CSLI
and the aerial surveillance at issue in this case. Leaders of a Beautiful Struggle, 2020 WL
1975380, at *12 (“Carpenter simply does not reach this case because CSLI offers a far
more intrusive, efficient, and reliable method of tracking a person's whereabouts than the
AIR pilot program.”). Whereas CSLI could be used to reliably track an individual’s
movement from day to day, AIR can only be used to track someone’s outdoor movements
for twelve hours at most. See Carpenter, 138 S. Ct. at 2220 (stating CSLI gives law
enforcement a “detailed chronicle of a person’s physical presence compiled every day,
every moment, over several years”). The technologies are also used quite differently. CSLI
is used by law enforcement to learn detailed information about someone it is already
monitoring. Id. at 2218 (explaining that CSLI tracking reveals an extensive amount of
private information to law enforcement). In contrast, AIR is used to identify suspects and
14

 witnesses to crimes; it takes no deep dive into an individual’s life and in fact can tell the
police very little about an identified person. Finally, Carpenter emphasized that CSLI
surveillance was “remarkably easy” and “cheap,” which meant that CSLI could easily be
used to track large numbers of identifiable people. Id. at 2217–18. AIR, in contrast, is not.
After flying a plane over the city, it takes an analyst hours to tag a person of interest and
reconstruct a couple of hours of his public movements.
Our opinion should not be overread. Although we conclude that AIR does not
invade a reasonable expectation of privacy, our decision should not be interpreted as
endorsing all forms of aerial surveillance. We only address the AIR program, which has
built-in limitations designed to minimize invasions of individual privacy. We do not
address a surveillance program that includes, for example, twenty-four hour surveillance
of indoor and outdoor spaces using photographs that allow analysts to immediately identify
the specific people being photographed. Although plaintiffs insist that AIR could spawn
something more intrusive than the program here, we must adjudicate each case based on
the facts before us. By proceeding cautiously, we heed Justice Frankfurter’s warning to
tread carefully in cases involving new technology, lest we “embarrass the future.” Nw.
Airlines, Inc. v. Minnesota, 322 U.S. 292, 300 (1944).
B.
In addition to not infringing a reasonable expectation of privacy, the AIR program
seeks to meet a serious law enforcement need without unduly burdening constitutional
rights. In such programmatic contexts, the Court assesses searches and seizures by
balancing the burden on constitutional rights against other law enforcement and public
15

 safety needs. See, e.g., Grady v. North Carolina, 575 U.S. 306, 310 (2015) (per curiam);
Samson v. California, 547 U.S. 843, 848 (2006) (“Whether a search is reasonable ‘is
determined by assessing, on the one hand, the degree to which it intrudes upon an
individual’s privacy and, on the other hand, the degree to which it is needed for the
promotion of legitimate governmental interests.’” (quoting United States v. Knights, 534
U.S. 112, 118–19 (2001)).
The Supreme Court has not given comprehensive guidance on when this balancing
test applies. It has, however, applied it in a wide variety of cases upholding programs and
policies addressing imperatives of public safety so long as—and this is crucial—the
program does not unreasonably burden personal privacy rights. See, e.g., King, 569 U.S. at
447 (upholding policy of swabbing the DNA of all arrestees for violent offenses); Bd. of
Educ. of Indep. Sch. Dist. No. 92 of Pottawatomie Cty. v. Earls, 536 U.S. 822 (2002)
(upholding program of drug testing of students who wished to participate in competitive
extracurricular activities); Nat’l Treasury Emps. Union v. Von Raab, 489 U.S. 656 (1989)
(holding that policy requiring employees seeking promotions to positions involving drug
enforcement to submit to urine tests was reasonable under balancing test). And in Michigan
Department of State Police v. Sitz, 496 U.S. 444 (1990), for example, the Supreme Court
upheld a program of drunk-driving checkpoints at which police officers randomly stopped
motorists without individualized suspicion. Id. at 447. The Court justified its decision, in
large part, by emphasizing the “magnitude” of the threat posed by drunk drivers to public
safety. Id. at 451 (“No one can seriously dispute the magnitude of the drunken driving
problem or the States’ interest in eradicating it.”).
16

 We emphasize that the Court has not just issued sweeping approvals of suspicionless
search programs. The key word is balancing. For example, the Court has repeatedly
reaffirmed that “searches made at the border, pursuant to the longstanding right of the
sovereign to protect itself by stopping and examining persons and property crossing into
this country, are reasonable simply by virtue of the fact that they occur at the
border . . . .” United States v. Ramsey, 431 U.S. 606, 616 (1977). But although the
government’s interest in conducting searches and seizures is high at the border, the Court
still balanced that law enforcement need against the privacy interests invaded. See, e.g.,
United States v. Flores-Montano, 541 U.S. 149, 154–56 (2004) (considering the privacy
interests of an individual whose car was partially disassembled during a border search).
And it recognized that a search could be intrusive enough or so destructive of property so
as to make it constitutionally unreasonable. Id. at 155–56.
Two factors support the validity of Baltimore’s AIR program under the Supreme
Court’s reasonableness balancing test for such programs. First, this program is not being
used to target particular individuals. Instead, it is used to track non-identified individuals
only under certain circumstances—namely, once a shooting, armed robbery, or carjacking
has been reported in a specific location. An individual will not have his public movements
tracked unless he happens to be at the scene of one of these violent crimes. Ignoring this
dynamic, plaintiffs complain they are particularly likely to be targeted because of their
community activism. But the AIR program cannot be used to target them as individuals;
they do not appear in the photographs as individuals, but as featureless pixelated dots.

17

 Second, the AIR program is designed to serve critical government purposes. Von
Raab, 489 U.S. at 665–66. Baltimore is plagued by violent crime, and the BPD has
struggled to capture the perpetrators. Although the BPD can take advantage of existing
networks of cameras and license plate readers, holes in their coverage may leave the police
dependent on eyewitnesses, who may not exist or be feasible to find. The AIR program is
designed to assist in that effort.
Baltimore’s AIR program thus passes the Supreme Court’s balancing test. On the
one hand, the BPD has a clearly demonstrated need for this surveillance. The violent crime
rates in Baltimore are astonishing, being indisputably among the worst in the country. And
despite law enforcement’s best efforts, its low clearance rate—just 32.1% in 2019 for
murders—shows the challenge is formidable and this additional tool important.
On the other hand, the program has been carefully designed to impose a minimal
burden on constitutional rights. We stress yet again that the BPD’s contract with PSS
establishes important limitations on how the AIR program can be used. Individuals
photographed by AIR are, as noted, represented as mere dots without race, gender, or other
identifiable characteristics. AIR only captures outdoor movements, and individuals cannot
be tracked once they enter a building. Analysts only access photographs when a violent
crime has occurred in a particular area. Unless photographs are being used in a prosecution,
they are discarded after forty-five days.
Moreover, this program burdens privacy substantially less than a well-established
staple of existing surveillance: security cameras. “Cameras mounted in public and semipublic places . . . are increasingly unremarkable, their presence taken for granted.” David
18

 Alan Sklansky, Too Much Information: How Not to Think About Privacy and the Fourth
Amendment, 102 CALIF. L. REV. 1069, 1085 (2014). Especially in large cities, these
cameras are ubiquitous, allowing “law enforcement to search for any individual, anywhere
in a city, going back for weeks or months . . . .” Rebecca Lipman, Protecting Privacy with
Fourth Amendment Use Restrictions, 25 GEO. MASON L. REV. 408, 436–37 (2018).
Baltimore itself has around eight hundred ground-based surveillance cameras. Our point is
not to endorse this or that use of such cameras, but to note that the AIR program is less
intrusive by comparison.
Precedent suggests law enforcement can use security cameras without violating the
Fourth Amendment. In Carpenter, the Supreme Court explicitly said it was not calling
these cameras into constitutional question. Carpenter, 138 S. Ct. at 2220. The lower courts
have generally upheld the use of security cameras against Fourth Amendment challenges.
See United States v. Vankesteren, 553 F.3d 286 (4th Cir. 2009); United States v. Houston,
813 F.3d 282 (6th Cir. 2016); United States v. Bucci, 582 F.3d 108 (1st Cir. 2009).
Assuming that these cameras are constitutional, as the Supreme Court has
instructed, it follows that the aerial surveillance in this case is too. The aerial cameras serve
a similar purpose as ground-based cameras: capturing public movements. Admittedly, they
are more effective at doing this than ground-based cameras because there are gaps in the
coverage of ground-based cameras. But as Judge Bennett recognized, AIR is less invasive
in most ways. Leaders of a Beautiful Struggle, 2020 WL 1975380 at *11 (“The AIR pilot
program does not approach the surveillance capabilities of a pole camera.”). Often
employing facial recognition software, ground-based cameras capture the individual
19

 characteristics of those surveilled, allowing the police to immediately conclude who
someone is and what they look like. See Rachel Levinson-Waldman, Hiding in Plain Sight:
A Fourth Amendment Framework for Analyzing Government Surveillance in Public, 66
EMORY L.J. 527, 539–41 (2017). AIR’s surveillance cannot do that. Indeed, it is telling
that the BPD cannot identify someone photographed by AIR without relying on groundbased cameras.
Rapid advancements in technology hold real promise in preventing and resolving
crimes. But they also pose unprecedented threats to personal privacy. Sometimes legitimate
concerns manage to meet at a sensible mean. So it is here. We agree with plaintiffs that
there are aerial surveillance programs that would transgress basic Fourth Amendment
protections. We further agree that investigative tools, whether aerial or electronic, should
not operate without restrictions. See, e.g., Omnibus Crime Control and Safe Streets Act of
1968, 18 U.S.C. §§ 2510-22 (establishing safeguards to regulate wiretapping). But we also
agree with the district court that this particular limited investigative tool does not violate
the Constitution. Finally, any search or seizure of a car, home, or person that follows from
an AIR investigation must of course satisfy the applicable probable cause and warrant
requirements expressly embedded in the Fourth Amendment.
V.
Plaintiffs also allege that AIR will violate their First Amendment rights to freely
associate with others. Although they did not extensively develop this argument below, their
basic theory is that the fear of being surveilled will deter them from associating with others
on the streets of Baltimore.
20

 We conclude plaintiffs are unlikely to succeed on the merits of this claim. The
Supreme Court has explained that the First Amendment protects two types of free
association: intimate association and expressive association. Roberts v. U.S. Jaycees, 468
U.S. 609, 619 (1984). Plaintiffs do not plead that either type of association is infringed
here. The AIR program does not in any way affect the personal affiliations that “attend the
creation and sustenance of a family,” such as marriage, childbirth, “the raising and
education of children,” or “cohabitation with one’s relatives.” Id. Nor can plaintiffs assert
that the AIR program interferes with their right “to speak, to worship, [or] to petition the
government for the redress of grievances.” Id. at 622.
The basic problem with plaintiffs’ argument is that people do not have a right to
avoid being seen in public places. And even if that were not so, it is a stretch to suggest
people are deterred from associating with each other because they may show up as a dot
under the AIR program. We thus conclude plaintiffs have failed to prove they will likely
succeed on the merits of their First Amendment claim.
VI.
In sum, the district court exercised sound equitable discretion in denying a
preliminary injunction. Allowing Baltimore’s AIR program to continue is the equitable
course of action and serves the public interest.
In deciding whether to grant a preliminary injunction, district courts “‘should pay
particular regard for the public consequences in employing the extraordinary remedy of
injunction.’” Winter, 555 U.S. at 24 (quoting Romero-Barcelo, 456 U.S. at 312). In Winter,
the Supreme Court explained that a preliminary injunction under the environmental laws
21

 against Navy training exercises was inappropriate because of the public interest in national
security and safety. Id. at 24–25.
The equitable balance here favors Baltimore. After a controversial initial roll-out of
the AIR program, the BPD pulled it back and consulted with the people of Baltimore. In
building political support for the program, the BPD agreed to the important limitations on
AIR’s capabilities that we have repeatedly underscored. Baltimore’s multiple partners in
the AIR program depict a community initiative in the process of ongoing evaluation. To
respond to such an initiative with an immediate “No” risks leaving a fine city with lessened
hope and not knowing where to turn.
In United States v. Jones, four justices of the Supreme Court recognized that elected
officials should play a leading role in crafting policies that balance the need for public
safety and the need for privacy. Jones, 565 U.S. at 429–30 (Alito, J., joined by Ginsburg,
Breyer, & Kagan, JJ., concurring in the judgment) (“A legislative body is well situated to
gauge changing public attitudes, to draw detailed lines, and to balance privacy and public
safety in a comprehensive way.”). That is exactly what Baltimore’s government officials
have done in this case. After the program’s secret rollout raised legitimate privacy
concerns, the program was halted. Constituents were heard and the BPD’s contract with
PSS was approved by elected officials with modified and substantially less intrusive
conditions.
Baltimore and its citizens have a strong interest in the AIR program not being
strangled in its infancy. The AIR program is carefully designed to slow the recurrent
increases in violent crime in Baltimore. As noted above, Baltimore is one of the more
22

 dangerous cities in America, with hundreds of people being murdered every year. Alec
MacGillis, The Tragedy of Baltimore, N.Y. Times Magazine, Mar. 17, 2019, at 32. But lest
we be swept up in harrowing statistics, remember that 348—the number of Baltimore
homicide victims in 2019—is not merely a number. These were 348 individual people,
with family and friends, hopes and dreams. Such a tragedy cannot be subsumed under some
new normal of indifference, and the public interest, see Winter, 555 U.S. at 24, is surely
not served by terminating at its very inception a program with the potential to help.
Plaintiffs have failed to persuade us that this program violates their constitutional
rights. As cherished as constitutional rights are, the Founders understood that the Bill of
Rights was not the only source of rights worthy of protection. They understood that all
people are endowed with “certain unalienable rights, that among these are Life, Liberty,
and the pursuit of Happiness.” The Declaration of Independence para. 2 (U.S. 1776). And
because these rights are “exposed to the invasion of others,” not just the government, the
Founders believed that government must exist for the “mutual preservation of [the
people’s] lives, liberties, and estates.” JOHN LOCKE, SECOND TREATISE ON GOVERNMENT
178 (Mark Goldie ed., 1993) (1689). Thus our Constitution was ratified, in no small part,
to “establish justice,” “insure domestic tranquility,” and “secure the blessings of liberty to
ourselves and our posterity.” U.S. Const. pmbl.
When hundreds of Baltimore residents are killed on their streets each year, their
rights to life are not protected. When murders remain unsolved, their rights to liberty are
not protected. When criminals can rob Baltimoreans at gunpoint with apparent impunity,
their rights to property are not protected. All these rights are denied without due process,
23

 nothing remotely resembling the protections an accused will receive, rightly and
deservedly, from the Bill of Rights. It is all so sad. Now and then the gunshots pause,
leaving only the silence of loved ones lost long before their time.
The AIR program just might help reduce this most aching of silences—giving it a
chance to succeed is in the public interest.
VII.
For the foregoing reasons, we hold that the district court did not abuse its discretion
in declining to issue the requested injunction.
AFFIRMED

24

 GREGORY, Chief Judge, dissenting:
This case presents a long-anticipated Fourth Amendment question:

Does the

Constitution permit warrantless dragnet surveillance by a police plane in the sky above an
American city? Until now, this question was merely a provocative hypothetical for law
professors 1 or a slippery-slope concern for litigants. 2 But with the Aerial Investigative
Research (“AIR”) program, Persistent Surveillance Systems (“PSS”) and the Baltimore
Police Department (“BPD”) bring the thought-exercise to life. Because the majority
concludes the warrantless surveillance program is constitutional, I respectfully dissent.

I.
I agree with the majority’s standing analysis. Maj. Op. at 6–7. I also broadly agree
with the majority’s presentation of the Fourth Amendment framework that applies here.

1

See, e.g., Rachel Levinson-Waldman, Hiding in Plain Sight: A Fourth Amendment
Framework for Analyzing Government Surveillance in Public, 66 Emory L.J. 527, 528
(2017) (“The 2002 movie Minority Report, which seemed wildly futuristic at the time,
effectively predicted many of the technologies now available to police at the click of a
button . . . . [D]oes an eye in the sky . . . constitute a search under the Fourth Amendment
. . . when it is used for public surveillance?”); Marc Jonathan Blitz, The Fourth Amendment
Future of Public Surveillance: Remote Recording and Other Searches in Public Space, 63
Am. Univ. L. Rev. 21, 24 (2013) (“Do police need to obtain a warrant . . . before they use
a drone to track a person’s movements or reconstruct those movements using video footage
from public cameras?”).
2

For example, in 1983, the Supreme Court considered a respondent’s fear that “the
result of the holding sought by the government would be that twenty-four hour surveillance
of any citizen of this country will be possible, without judicial knowledge or supervision.”
United States v. Knotts, 460 U.S. 276, 283 (1983) (internal quotations omitted). The Court
was skeptical, but noted that “if such dragnet type law enforcement practices as respondent
envisions should eventually occur, there will be time enough then to determine whether
different constitutional principles may be applicable.” Id. at 283–84.
25

 Maj. Op. at 9–10. We apply the reasonable expectation of privacy test, which has a
subjective and objective component. United States v. Castellanos, 716 F.3d 828, 832 (4th
Cir. 2013). First, we look to whether an individual’s subjective expectation of privacy is
implicated; then, we ask whether that expectation is objectively reasonable. Id.
Our minds diverge, however, when applying this test to the AIR program. See Maj.
Op. 10–20. The majority finds—as the district court found below—that no reasonable
expectation of privacy is presented here because the AIR program is capable of only
short-term surveillance. Maj. Op. at 10–15. Therefore, they reason, this case is more like
those cases that have approved of public surveillance tactics—like ground-based security
cameras or aerial “flyover” surveillance—than those that have struck down more revealing
tracking. On this basis, the majority distinguishes the Supreme Court’s recent teachings in
Carpenter v. United States, 138 S. Ct. 2206 (2018). Alternatively, the majority also finds
that the AIR program is justified as a “programmatic search.” Maj. Op. at 15–20.
These conclusions rest on a fundamentally warped understanding of the facts,
accepting the Government’s promises about the AIR program and ignoring the plaintiffs’
contrary evidence. The AIR program does, indeed, amount to long-term surveillance that
compiles “a detailed and comprehensive record” of a person’s past movements. See
Carpenter, 138 S. Ct. at 2215–19. Thus, Carpenter is not only relevant to this case. It
controls the outcome.
A.
Carpenter clarified the rules governing law enforcement surveillance of people’s
public movements. Id. at 2215–16 (internal quotations omitted). The starting point was
26

 United States v. Knotts, 460 U.S. 276 (1983), where the Court held that the use of a
location-tracking “beeper” on a suspect’s car was constitutional because a person
“traveling on public thoroughfares has no reasonable expectation of privacy in [their]
movements from one place to another,” given that they “voluntarily conveyed [their
movements] to anyone who wanted to look.” Id. at 281–82. The Knotts court qualified,
though, that if “dragnet type law enforcement practices . . . should eventually occur,” then
“different constitutional principles may be applicable.” Id. at 284.
“Three decades later, the Court considered more sophisticated surveillance of the
sort envisioned in Knotts and found that different principles did indeed apply.” Carpenter,
138 S. Ct. at 2215. In United States v. Jones, 565 U.S. 400 (2012), police used a GPS
tracking device to “remotely monitor[] [a] vehicle’s movements for 28 days.” Carpenter,
138 S. Ct. at 2215. Though Jones ultimately rested on trespass principles, it also had
privacy implications, as Carpenter explained: “Since GPS monitoring of a vehicle tracks
‘every movement’ a person makes in that vehicle, the [five] concurring Justices concluded
that ‘longer term GPS monitoring . . . impinges on expectations of privacy’—regardless
whether those movements were disclosed to the public at large.” Id. (quoting Jones, 565
U.S. at 430 (Alito, J., concurring); Jones, 565 U.S. at 415 (Sotomayor, J., concurring)).
Thus, “[a] person does not surrender all Fourth Amendment protection by venturing
into the public sphere.” Id. at 2217. Carpenter solidified the line between permissible
short-term tracking of a person in public—just as law enforcement was capable of
accomplishing “[p]rior to the digital age”—and an individual’s “expectation of privacy in
the whole of their physical movements.” Id. For example, while most people understand
27

 that a security camera may record them as they walk down a city block, they also
reasonably expect that police will not “secretly monitor and catalogue every single
movement of an individual’s car for a very long period.” See id. (quoting Jones, 565 U.S.
at 430 (Alito, J., concurring)).
Based on this rule, the Court held that the Fourth Amendment protects cell-site
location information (“CSLI”) because, “[m]uch like GPS tracking of a vehicle, cell phone
location information is detailed, encyclopedic, and effortlessly compiled.” Id. at 2216–19.
CSLI reveals “an all-encompassing record of the holder’s whereabouts,” and “[a]s with
GPS information, the time-stamped data provides an intimate window into a person’s life,
revealing not only [their] particular movements, but through them [their] ‘familial,
political, professional, religious, and sexual associations.’” Id. at 2217 (quoting Jones, 565
U.S. at 415 (Sotomayor, J., concurring)). Such a record “hold[s] for many Americans the
privacies of life.” Id. (quoting Riley v. California, 573 U.S. 373, 403 (2014)). Even more
than a GPS device on a car, cell data “tracks nearly exactly the movements of its owner,”
such that “when the Government tracks the location of a cell phone, it achieves near perfect
surveillance, as if it had attached an ankle monitor to the phone’s user.” Id. at 2218. And,
the “retrospective quality” of cell data further distinguishes it from past surveillance
techniques, because “the Government can now travel back in time to retrace a person’s
whereabouts”; “[u]nlike with the GPS device in Jones, police need not even know in
advance whether they want to follow a particular individual, or when.” Id. (“Whoever the
suspect turns out to be, [they] ha[ve] effectively been tailed every moment of every day . . .

28

 and the police may—in the Government’s view—call upon the results of that surveillance
without regard to the constraints of the Fourth Amendment.”).
Carpenter’s application here could not be clearer. The AIR program compiles a
“detailed, encyclopedic” record of “the whole of” every resident of Baltimore’s movements
in public. See id. at 2216–19; see also id. at 2217 (“Whether the Government employs its
own surveillance technology as in Jones or leverages the technology of a wireless carrier,
we hold that an individual maintains a legitimate expectation of privacy in the record of
[their] physical movements . . .”) (emphasis added).
Carpenter identified Jones as applying new constitutional principles based on the
possibility of dragnet surveillance anticipated in Knotts. Id. at 2215. Here, more like the
cell phone data in Carpenter and GPS-tracking in Jones than the beeper-tracking in Knotts,
the AIR program “tracks every movement” of every person outside in Baltimore. See id.
And, like the cell data in Carpenter, the AIR program’s privacy implications are amplified
because of the “retrospective quality” of its data; within a 45-day window, 3 the Baltimore
police can “travel back in time to retrace a person’s whereabouts.” See id. at 2218.

3

Notably, though, the 45-day retention period is self-imposed by BPD and PSS.
The data could theoretically be retained for longer periods or indefinitely. For example,
when BPD initially, and secretly, piloted the AIR program in 2016, its data was retained
indefinitely—even though there was, then too, a purported 45-day retention period. See
Nat’l Police Found., Review of the Baltimore Police Department’s Use of Persistent
Surveillance 18 (2017), https://perma.cc/Q277-QP8U (finding, after initial pilot of the AIR
program, that “[t]he PSS Privacy Protection Policy states that after 45 days, the data will
be destroyed unless an inquiry is received. However, PSS is retaining all the data from the
. . . pilot program”) (saved as ECF opinion attachment).
29

 This capability transcends the expectation that was at issue in Knotts—that police
officers could follow a person in public—and instead “gives police access to a category of
information otherwise unknowable.” See id. Just like in Carpenter, the AIR program
means that for every single resident of Baltimore, “[w]hoever the suspect turns out to be,
[they] ha[ve] effectively been tailed every moment of every day” for 45 days. See id.
Residents of Baltimore have a subjective expectation of privacy in the whole of these
day-to-day movements, even though they take place outdoors. See id. at 2215–19; Jones,
565 U.S. at 415 (Sotomayor, J., concurring); Jones, 565 U.S. at 430 (Alito, J., concurring).
Long-term, recorded surveillance of public movements uncovers more than
temporary trailing by a suspecting officer; it reveals a person’s most intimate associations
and activities. Carpenter, 138 S. Ct. at 2217–18. Therefore, that subjective expectation of
privacy is objectively reasonable because it concerns the “privacies of life”—the epitome
of information expected to be beyond the warrantless reach of the government. See id. at
2214, 2218; Boyd v. United States, 116 U.S. 616, 630 (1886) (explaining that the
fundamental principles of the Fourth Amendment protect “the sanctity of a [person]’s home
and the privacies of life”). Accordingly, when police call upon AIR program data, the
“information obtained . . . [is] the product of a search,” and the program’s warrantless
operation violates the Fourth Amendment. Carpenter, 138 S. Ct. at 2217; see also Jones,
565 U.S. at 415 (Sotomayor, J., concurring) (questioning “the appropriateness of entrusting
to the Executive, in the absence of any oversight from a coordinate branch, a tool so
amenable to misuse, especially in light of the Fourth Amendment’s goal to curb arbitrary

30

 exercises of police power and prevent ‘a too permeating police surveillance’”) (quoting
United States v. Di Re, 332 U.S. 581, 595 (1948)).
B.
The majority distinguishes Carpenter by finding that the AIR program is incapable
of long-term and individually-targeted surveillance. On that basis, the majority concludes
that plaintiffs have no reasonable expectation of privacy in the information the AIR
program obtains.

The majority also finds, alternatively, that the AIR program is

constitutional as a “programmatic search.”
1.
The majority, the district court, and the Government each emphasize certain aspects
of the AIR program to distinguish Carpenter and Jones. In particular, the majority relies
on the fact that the surveillance planes fly for “only” twelve hours per day, and that
individuals, when photographed, appear “only” as “pixelated dots.” See Maj. Op. at 4–5.
Indeed, the majority emphasizes this latter fact repeatedly throughout its opinion. See, e.g.,
Maj. Op. at 3–6, 12, 17, 18, 21.
The majority derives far-reaching conclusions from these two facts.

First,

“[b]ecause [the planes] do not fly at night, AIR surveillance cannot be used to track
individuals from day-to-day.” Maj. Op. at 11. Second, because people appear as “pixelated
dots,” “individuals cannot be tracked once they enter a building,” and “the AIR program
cannot be used to target them as individuals.” Maj. Op. at 17–18. Thus, the AIR program
is “less intrusive” than ground-based security cameras, which also “captur[e] public
movements”—with significantly more detail, though “[a]dmittedly” with “gaps in the
31

 coverage”—and remain constitutional after Carpenter. Maj. Op. 18–20. By conceiving of
the AIR program as incapable of identifying individuals or tracking them over time, the
majority finds Carpenter inapplicable. See Maj. Op. at 11–20.
However, these conclusions only hold up when limiting the Fourth Amendment
analysis of the AIR program to solely the photographic data its planes collect, as opposed
to what information that data can reveal. Carpenter precludes such a narrow focus. The
Carpenter court did not consider only what was revealed by raw CSLI, but also what
information law enforcement could deduce “from the . . . location data it received . . . in
combination with other information.” 4 138 S. Ct. at 2218 (emphasis added). The Court
rejected the Government’s argument that using the CSLI was not a search because “the
location records did not on their own suffice to place Carpenter at the crime scene.” Id.
(internal quotations and alterations omitted). Rather, because the CSLI enabled the police
to “deduce a detailed log of Carpenter’s movements,” a warrant was required. See id. at
2217–18. Thus, the rule from Carpenter is that if locational data enables police to deduce
a cumulative, retrospective “record of [a person’s] physical movements,” then “[t]he
location information obtained . . . was the product of a search.” See id.

4

In Carpenter, the records themselves were only a time-stamped log of cell tower
locations pinged by the suspect’s cell phone number. See Carpenter, 138 S. Ct. at 2212–
13. Police then used that data to “produce[] maps” that “placed Carpenter’s phone near
four of the charged robberies.” Id. Here, as in Carpenter, the police get raw locational
data without a warrant and then apply policework to track and identify suspects using that
data, just like producing maps of the Carpenter suspect’s likely movements. See, e.g.,
Kyllo v. United States, 533 U.S. 27, 36–37 (2001) (explaining that, where police used
technology to warrantlessly infer information that violated reasonable expectations of
privacy, it was “[t]he police activity” that “was held to be a search, and the search was held
unlawful”) (emphasis added).
32

 This Court must apply the same scope of analysis here. While aerial photographs
and ground-based security camera footage both capture people’s public movements, the
plaintiffs challenge the AIR program specifically for capturing images of everyone, every
day, and everywhere. It makes no Fourth Amendment difference that the AIR photographs
may require further analysis to yield their investigatory value. See id. at 2218 (rejecting
“the proposition that ‘inference insulates a search’”) (quoting Kyllo v. United States, 533
U.S. 27, 37 (2001)). So while the majority posits that AIR photographs “take[] no deep
dive into an individual’s life and in fact can tell the police very little about an identified
person,” Maj. Op. at 15, the question under Carpenter is whether the AIR program will
nevertheless enable police to deduce a comprehensive record of people’s past movements.
It is that capability that distinguishes CCTV cameras, or the temporary surveillance at issue
in Knotts, and tells the police more than a little about an identified person. See Jones, 565
U.S. at 415 (Sotomayor, J., concurring) (explaining how retrospective records of public
movements can reveal “a wealth of detail” about a person’s “familial, political,
professional, religious, and sexual associations” and private activities like trips to a
psychiatrist, plastic surgeon, abortion clinic, AIDS treatment center, and more).
Applying that broader perspective to this case, the limitations of the AIR program
that the majority emphasizes do little to distinguish Carpenter and Jones. Indeed, it is hard
to reconcile the majority’s portrayal of the AIR program’s capabilities with its insistence
that the program is justified because it helps BPD police crime. That the AIR program’s
surveillance planes will fly only during the daylight hours and capture individuals as

33

 solitary pixels 5 does not mean that AIR program data cannot be used to track specific
individuals over time. Presumably, the vast majority of Baltimore’s residents start and end
most days in the same place: their homes. 6 Who resides at a given address is often public
information. Likewise, many people begin each day at home, then travel during the
daytime based on a daily routine, and then return home again. In just one possible
application, law enforcement could use AIR data to track a person’s movements from a
crime scene to, eventually, a residential location where the person remains. Police could
then look through time and track the movements of people from that residence. Police
could use any number of context clues to distinguish individuals and deduce identity. After
all, that is the very purpose of the program.
The plaintiffs submitted evidence to the district court to support this intuitive
reasoning. In a study based on cell phone locational data, researchers found that a person’s
movements—looked at as a whole—tend to be so unique that 95% of people can be

5

Both of these limitations are self-imposed by BPD and PSS, not technological
necessities. The plane-cameras are capable of taking decipherable photographs of city
streets lit by ambient light during the nighttime. Likewise, BPD and PSS “agreed” to set
the resolution of the cameras at “one pixel per person,” but “the technology has the ability
to upgrade the quality.” See BPD, Aerial Investigations Research Pilot Program
Presentation
(AIR)
at
23:46,
Facebook
(Mar.
30,
2020),
https://www.facebook.com/58771761955/videos/212014970074066 (“BPD Facebook
Live Townhall”) (saved as ECF opinion attachment).
6

C.f. U.S. Census Bureau, American Community Survey, 2018 American
Community Survey 1-Year Estimates, Table B25033 (Baltimore City),
https://perma.cc/CM9C-VC95 (estimating, among population living in occupied housing
units, that over 75% of Baltimore City residents reside in single-unit housing structures)
(sum of persons in owner-occupied and renter-occupied single unit structures, mobile
homes, and “boats, RVs, van, etc.” divided by total population) (saved as ECF opinion
attachment).
34

 individually identified based only on “[f]our randomly chosen points” of location data from
their collective movements. 7 In other words, when reviewing data showing a group of
people’s collective movements, an individual’s movements tend to be so unique that it is
not difficult to distinguish among people in the group. This is why the majority’s
conclusion that individuals cannot be tracked is unsubstantiated. The police would have
access to 45 daytimes’ worth of retroactive locational data to follow any given “pixel” as
it moved through time and space. By analyzing patterns, it is possible—and often relatively
easy—to identify the person behind the pixel based on the routines it follows.
And that is without even considering BPD’s ability to rely on other police
information systems. BPD’s contract with PSS expressly provides for integrating BPD’s
“CitiWatch CCTV cameras” and “License Plate Readers” into the AIR program, stating
that “analysts will track individuals and vehicles” using both systems, “to provide more
detailed descriptions” and “allowing the vehicles to be identified.” BPD’s “CitiWatch”
surveillance camera network consists of more than 800 cameras in Baltimore City. 8 Along
with downtown Baltimore, those cameras are most prevalent in predominantly Black

7

The district court ignored the study by finding that the plaintiffs did not “proffer[]
any evidence suggesting that the same analysis applicable to cell-phone location data may
be grafted on to the imagery data produced by the AIR pilot program.” But no such
evidence is necessary; the source of the location data is irrelevant to the findings. The
study establishes the relative ease of “re-identifying” unique individuals from a pool of
data showing their collective movements. It would make no difference whether that pool
was sourced from cell phones, GPS trackers, surveillance airplanes, or some other source.
8

BPD Facebook Live Townhall at 28:07.
35

 neighborhoods. 9 Thus, considering all of the above, the fact that AIR data does not include
nighttime and indoor movements does not inherently inhibit “track[ing] individuals from
day-to-day.”
According to the majority, “that the police may be able to use . . . security cameras
and license plate readers[] in conjunction with AIR photographs” is beyond this Court’s
review because “plaintiffs do not challenge these existing tools—only the AIR program in
particular.” Maj. Op. at 11–12. But as discussed above, there is no basis for this insistence
that the Court put on blinders when considering how the AIR program is being
implemented. 10 That plaintiffs do not independently challenge the legality of surveillance
cameras and license plate readers does not mean that this Court must ignore their

9

Compare City of Baltimore, Open Baltimore: CCTV Cameras,
https://data.baltimorecity.gov/Public-Safety/CCTV-Cameras/y3f4-umna (showing map of
locations of CitiWatch CCTV cameras) (saved as ECF opinion attachment), with U.S.
Dep’t of Justice, Civil Rights Div., Investigation of the Baltimore City Police Department
13 (2016), https://perma.cc/EPN2-7G38 (showing map of Black population in Baltimore
City) (saved as ECF opinion attachment).
10

There is no question that police “may be able to use” these other surveillance
systems; their integration is expressly provided for in BPD’s contract with PSS and is
fundamental to the operation of the AIR program. PSS provided that it will “hire between
15 and 25 analysts for this Project,” working “in two shifts per day, seven days per week
in support of this effort.” And, the contract provides that PSS will integrate BPD’s existing
surveillance systems—Computer Aided Dispatch, CitiWatch Ground Based Cameras
Integration, Shot Spotter Gun Shot Detection System, and License Plate Readers—into its
own. Within 18 hours of a “Target Crime,” PSS provides that it will analyze the scene of
the crime, identify people and vehicles present, and then analyze “locations the vehicles
and people visited prior to and after the crime,” as well as the “tracks of people and vehicles
that met with people who were tracked from the crime scene and the locations they came
from and went to,” by using, in addition to PSS’ own photographic data, “ground-based
camera video made available to [PSS] by BPD.”
36

 availability to the police and integration in the AIR program when applying Carpenter.
See 138 S. Ct. at 2217–18.
Likewise, the twelve-hours-per-day, single-pixel nature of the AIR program’s data
hardly distinguishes Carpenter and Jones. The CSLI at issue in Carpenter was not a
comprehensive location log without any informational gaps. As the lower court explained,
each datapoint covered an area “typically anywhere from a half-mile to two miles.” United
States v. Carpenter, 819 F.3d 880, 885 (6th Cir. 2016). And datapoints were generated
only “at call origination and at call termination for incoming and outgoing calls.” See id.
at 884. Similarly, in Jones, police collected 28 days of data showing the movements of the
suspect’s car “within 50 to 100 feet.” 565 U.S. at 403. This kind of GPS-data would only
reflect a suspect’s movements in their own car and could only place a suspect in the vicinity
of a given site where their car was parked, without indicating, for example, which building
the suspect may have entered. See Carpenter, 138 S. Ct. at 2218. That the AIR data has
gaps overnight or when a suspect is indoors is no different from the gaps in Carpenter
between phone calls or in Jones between car rides. 11 And the AIR photographs provide a
more precise depiction of individuals’ locations than the data in either Carpenter or Jones.

11

It is worth mentioning, though, that even if we were to consider “only” the AIR
program’s single twelve-hour daily recordings, there is still likely a Fourth Amendment
violation. The most problematic aspect of the AIR program is its dragnet nature,
surveilling every individual’s public movements across Baltimore. Carpenter emphasized
this concern regarding CSLI. 138 S. Ct. at 2218 (“Critically, because location information
is continually logged for all of the 400 million devices in the United States—not just those
belonging to persons who might happen to come under investigation—this newfound
tracking capacity runs against everyone.”); see also Jones, 565 U.S. at 415–16 (noting that
developing technology means that “even short-term monitoring” that obtains “a precise,
comprehensive record of a person’s public movements” can reveal private associations).
37

 Ultimately, the majority’s portrayal of the AIR program aligns with BPD’s public
assurances about how the program is intended to be used, but requires putting one’s head
in the sand about the program’s capabilities. That latter capability, however, is what
controls the Fourth Amendment analysis. See id. at 2219 (noting that the CSLI at issue
“reflect[ed] the state of technology at the start of the decade,” but considering “new
technology” for CSLI data that is even more precise); Kyllo, 533 U.S. at 35–36 (rejecting
“a mechanical interpretation of the Fourth Amendment” that would leave protections “at
the mercy of advancing technology,” and instead instructing that “the rule we adopt must
take account of more sophisticated systems that are already in use or in development”).
2.
Once the assumption that the AIR program cannot identify and track individuals
over time is dispelled, the majority’s corresponding application of the law breaks down.
First, conceiving of the program as short-term surveillance, the majority, like the district
court, concludes that this case is governed by Knotts and its progeny, analogizing to the
line of cases approving of “various types of aerial surveillance . . . more intrusive than that
in AIR.” Maj. Op. at 11–13. The Supreme Court has approved of “flying a plane a mere
one thousand feet over an individual’s home to photograph items within a fenced-in
backyard”; using a plane flying over a factory to “photograph[] items as small as one half
inch in diameter”; and circling a helicopter “four hundred feet above a home . . . to look
into a greenhouse partially within the home’s curtilage.” Maj. Op. at 12–13. By contrast,
“the AIR cameras cannot photograph an individual item within a backyard” and “AIR
involves no surveillance of an individual’s home or curtilage,” so the program is “decidedly
38

 less intrusive.” Maj. Op. at 13. “Individuals are just dots, and analysts cannot identify a
person’s race or sex from an AIR photograph,” such that “AIR captures no ‘intimate
details’ of the sort that might implicate Fourth Amendment protections.” Maj. Op. at 12.
This reasoning is nonresponsive to the Fourth Amendment objections in this case.
Plaintiffs do not challenge the AIR program on the basis of what any single photograph
may reveal. Rather, they object to the planes photographing the city for twelve hours per
day, seven days per week, and creating a retrospective database of everyone’s movements
across the city. All of the cases the majority cites, and that the district court relied on,
involved single instances of targeted surveillance of an individual suspect or location for a
temporary period of time. 12 Because that kind of surveillance can only reveal what is
already being held out in the open and could be observed by traditional surveillance
tactics—albeit, sometimes with the aid of “augmented” sensory perception—it is
permissible under the Fourth Amendment. See Carpenter, 138 S. Ct. at 2215 (citing

12

See California v. Ciraolo, 476 U.S. 207, 209, 213–14 (1986) (describing single
occasion of “fl[ying] over respondent’s house at . . . 1,000 feet, within navigable airspace”
to “observe plants” with “the naked eye”); Dow Chem. Co. v. United States, 476 U.S. 227,
229, 238 (1986) (approving of “commercial aerial photographer” flying over a suspected
industrial building without using “highly sophisticated surveillance equipment not
generally available to the public”); Florida v. Riley, 488 U.S. 445, 449 (1989) (finding no
violation where officer “circled twice over respondent’s property in a helicopter” at 400
feet to observe greenhouse “[w]ith his naked eye”); United States v. Breza, 308 F.3d 430,
433–34 (4th Cir. 2002) (finding no violation when, “[w]hile conducting a routine drug
interdiction operation,” officers “flew over [the suspect’s] property in a helicopter” to
observe marijuana plants); Giancola v. W.Va. Dep’t of Pub. Safety, 830 F.2d 547, 548,
550–51 (4th Cir. 1987) (looking to factors such as “the total number of instances of
surveillance, the frequency of surveillance, [and] the length of each surveillance,” and
finding no violation when “two helicopters . . . flew over [the suspect’s] property for ten to
twenty minutes” to observe fields in plain view).
39

 Knotts, 460 U.S. at 281–82). But the ability to call upon a log of photographs revealing an
individual’s movements in the past is not just augmented sensory perception; it “gives
police access to a category of information otherwise unknowable.” See id. at 2218.
Therefore, that information is the product of a search. See id. at 2217.
The AIR program reveals intimate details about the plaintiffs’ lives because it
records where they go, not because any single photograph in isolation is especially
revealing. The fact that the database of past movements comes from aerial surveillance
planes instead of, say, a cell-service provider’s CSLI records is, ultimately,
inconsequential. Thus, despite their discussion of flying police technology, the cases the
district court relied on and the majority points to bear little relevance to this case.
3.
Finally, the majority also considers the AIR program’s constitutionality as a
“programmatic search,” applying the balancing test that applies to drunk driving
checkpoints, searches at border crossings, and searches in carceral settings. Maj. Op. 15–
20. These cases are simply inapplicable. Indeed, the parties did not brief the issue because
the Government did not argue that the AIR program is governed by the doctrine.
Nevertheless, the majority reasons that it provides an alternative basis for affirming.
What the majority refers to as programmatic searches are one type of search that
“fit[s] within the closely guarded category of constitutionally permissible suspicionless
searches.” See Chandler v. Miller, 520 U.S. 305, 309 (1997); see generally United States
v. Curry, 965 F.3d 313, 339–43 (4th Cir. 2020) (Diaz, J., concurring). “The suspicionless
search is the very evil the Fourth Amendment was intended to stamp out,” but
40

 individualized suspicion can be “dispensed with” as a component of Fourth Amendment
reasonableness when certain kinds of searches are necessary to meet a “special need” of
the Government. Samson v. California, 547 U.S. 843, 858 (2006) (Stevens, J., dissenting).
Thus, “[i]n limited circumstances, where the privacy interests implicated by the search are
minimal, and where an important governmental interest furthered by the intrusion would
be placed in jeopardy by a requirement of individualized suspicion, a search may be
reasonable despite the absence of such suspicion.” Skinner v. Ry. Labor Execs. Ass’n, 489
U.S. 602, 624 (1989).
In addition to this balancing requirement, the suspicionless search also must take an
especially “systematic” and “discretionless” form. See Curry, 965 F.3d at 339–43 (Diaz,
J., concurring). For example, this doctrine provides the authority for routinized searches
at a fixed Border Patrol checkpoint designed to intercept undocumented persons. Mich.
Dep’t of State Police v. Sitz, 496 U.S. 444, 448–55 (1990). Yet, the doctrine does not
authorize “roving-patrol stops,” even for the same governmental purpose, because of their
“sporadic and random” nature and the “unbridled discretion” they grant to officers.
Almeida-Sanchez v. United States, 413 U.S. 266, 270–75 (1973).
Here, the majority concludes that the AIR program is an acceptable suspicionless
programmatic search program for two reasons. The first is that “this program is not being
used to target particular individuals. Instead, it is used to track non-identified individuals
only under certain circumstances;” namely, when those individuals “happen[] to be at the
scene of . . . violent crimes.” Maj. Op. at 17. This proposition defies reason. In one breath,
the majority states that the AIR program is “not being used to target particular individuals,”
41

 while in the next, it specifies that the program “track[s] non-identified individuals” who
are present at crime scenes. The fact that police may not know the name of the individual
they are surveilling—at the time the surveillance initially takes place—does not change
the reality of what is occurring.

When police use the AIR program to track

as-yet-unidentified individuals, they are “target[ing] particular individuals.” And the
purpose of using the AIR program to track those non-identified individuals from crime
scenes is to identify them, be it as witnesses or as suspects. What other reason could there
be for investing millions of dollars in a city-wide aerial surveillance program?
Indeed, that is the very governmental purpose that the majority then provides as its
second reason for finding a constitutional suspicionless search here:
Second, the AIR program is designed to serve critical government purposes.
Baltimore is plagued by violent crime, and the BPD has struggled to capture
the perpetrators. Although the BPD can take advantage of existing networks
of cameras and license plate readers, holes in their coverage may leave the
police dependent on eyewitnesses, who may not exist or be feasible to find.
The AIR program is designed to assist in that effort.
Maj. Op. at 18 (internal citation omitted). If the AIR program is justified because it assists
in the effort of capturing the perpetrators of violent crime—by filling holes in police
surveillance coverage and providing eyewitness leads—then the AIR program is targeting
and identifying individual people. Such a program is clearly incompatible with the
“discretionless” and “systematic” suspicionless search rubric that the majority attempts to
force onto it.
The majority’s position is more comprehensible if one conceives of the
“programmatic search” as being solely the act of taking photographs of the city, as distinct

42

 from the act of reviewing those photographs to deduce an individual’s movements. But,
as discussed above, Carpenter precludes such a framing. Carpenter identified the “search”
as occurring “when the Government accessed CSLI from the wireless carriers,” because
“[f]rom the . . . location data it received, the Government could, in combination with other
information, deduce a detailed log of Carpenter’s movements.” 138 S. Ct. at 2218
(emphasis added). In other words, because the interest at stake is one’s expected privacy
“in the whole of [their] physical movements,” the conduct amounting to a “search” includes
the police activity that “deduce[s]” the “detailed log.” See id.; see also Kyllo, 533 U.S. at
36 (finding “the dissent’s extraordinary assertion that anything learned through ‘an
inference’ cannot be a search” is “blatantly contrary to United States v. Karo, where the
police ‘inferred’ from the activation of a beeper that a certain can of ether was in the home.
The police activity was held to be a search, and the search was held unlawful”) (emphasis
added).
Regardless, even if accepting the majority’s framing, the “programmatic search”
doctrine cannot render the AIR program constitutional. A foundational rule of the doctrine
is that the “special need” justifying the suspicionless search must be “beyond the normal
need for law enforcement.” Curry, 965 F.3d at 340 (Diaz, J., concurring); see, e.g.,
Ferguson v. Charleston, 532 U.S. 67, 79 (2001) (referring to whether the “special need” is
“divorced from the State’s general interest in law enforcement”); City of Indianapolis v.
Edmond, 531 U.S. 32, 37 (2000) (holding that “narcotics-detection” checkpoint lacked
sufficient “special need” because its “primary purpose was to detect evidence of ordinary
criminal wrongdoing”); Griffin v. Wisconsin, 483 U.S. 868, 873 (1987) (looking to whether
43

 “special needs, beyond the normal need for law enforcement, make the warrant and
probable-cause requirement impracticable”).
Here, the “critical government purpose” identified by the majority as the “special
need” is that “Baltimore is plagued by violent crime, and the BPD has struggled to capture
the perpetrators.” Stopping violent crime by capturing perpetrators is a paradigmatic
example of “the normal need for law enforcement.” See Edmond, 531 U.S. at 37–40.
Because the majority does not identify a special need other than “the general interest in
crime control” to justify the AIR program as a suspicionless search system—and the
program serves no other special purpose—the doctrine does not apply. See id. at 40.

II.
I therefore find that the plaintiffs established a strong likelihood of success on the
merits of their Fourth Amendment claim. The remaining preliminary injunction factors
counsel in favor of plaintiffs as well.
Because the plaintiffs identify a Fourth Amendment violation, they are irreparably
harmed. See, e.g., Ross v. Meese, 818 F.2d 1132, 1134–35 (4th Cir. 1987) (finding, in
Fourth Amendment context, that “the denial of a constitutional right, if denial is
established, constitutes irreparable harm for purposes of equitable jurisdiction”); see also
Elrod v. Burns, 427 U.S. 347, 373 (1976) (finding, where plaintiff established a likelihood
of success on First Amendment claim, that the violation “unquestionably constitutes
irreparable injury”); Preston v. Thompson, 589 F.2d 300, 303 n.4 (7th Cir. 1978) (“[A]
continuing constitutional violation constitutes proof of an irreparable harm.”).
44

 Likewise, the balance of the equities favors plaintiffs because they have established
an ongoing constitutional violation. By contrast, “precedent counsels that ‘a state is in no
way harmed by issuance of a preliminary injunction which prevents the state from
enforcing restrictions likely to be found unconstitutional.’”

See Centro Tepeyac v.

Montgomery County, 722 F.3d 184, 191 (4th Cir. 2013) (quoting Giovani Carandola Ltd.
v. Bason, 303 F.3d 507, 521 (4th Cir. 2002)) (“If anything, the system is improved by such
an injunction.”). Moreover, BPD has emphasized in its public-facing materials that the
present AIR program is only “a 180 day pilot program” designed to “collect enough data
that can be analyzed and reviewed” to determine the program’s effectiveness. BPD has
also represented that “if research determines little to no impact on the measures for success,
then BPD will ground the plane and discontinue operations.” That limited purpose would
be unimpeded if it takes place after this case is resolved, whereas plaintiffs’ injuries are
accruing every passing day.
Finally, for these same reasons, the public interest favors the plaintiffs. “Surely,
upholding constitutional rights serves the public interest.” Newsom ex rel. Newsom v.
Albemarle Cty. Sch. Bd., 354 F.3d 249, 261 (4th Cir. 2003); see also League of Women
Voters of N.C. v. North Carolina, 769 F.3d 224, 248 (4th Cir. 2014) (same); Giovani
Carandola, 303 F.3d at 521 (same). The Government’s desire to find innovative solutions
to respond to crime is also generally in the public interest, but not when that purpose comes
at the expense of the public’s constitutional rights.

45

 III.
Therefore, the district court abused its discretion in denying plaintiff’s request for a
preliminary injunction. The district court portrayed the AIR program’s capabilities as
unrealistically limited while ignoring countervailing evidence. And the district court relied
on precedents that bear little relevance to plaintiff’s Fourth Amendment objections to the
AIR program. Instead, Carpenter controls this case.
Throughout its opinion, the majority emphasizes that “Baltimore is one of the more
dangerous cities in America, with hundreds of people being murdered every year.” Maj.
Op. at 22–23. And BPD “has struggled to respond effectively.” Maj. Op. at 2. The
majority concludes that, because “[t]he AIR program just might help reduce” these levels
of violence, we should “giv[e] it a chance to succeed.” Maj. Op. at 24.
But dragnet law enforcement tactics are nothing new for many neighborhoods in
Baltimore. See generally Lucius T. Outlaw III, Unsecured (Black) Bodies: How Baltimore
Foreshadows the Dangers of Racially Targeted Dragnet Policing Let Loose by Utah v.
Strieff, 50 N.M. L. Rev. 25, 36–45 (2020). A Department of Justice report, following a
14-month investigation into BPD’s unconstitutional practices, provides several examples.
See U.S. Dep’t of Justice, Civil Rights Div., Investigation of the Baltimore City Police
Department (2016), https://perma.cc/EPN2-7G38 (“DOJ Report”) (saved as ECF opinion
attachment).
The DOJ found “a widespread pattern of BPD officers stopping and detaining
people on Baltimore streets without reasonable suspicion that they are involved in criminal
activity.” Id. at 27. BPD officers “routinely violate” the Fourth Amendment-based rules
46

 governing Terry-stops by detaining and questioning people who are simply “sitting,
standing, or walking in public areas, even where officers have no basis to suspect them of
wrongdoing.” Id. at 28. And “overwhelming” statistical evidence established that these
strategies targeted predominantly Black neighborhoods and “produce severe and
unjustified disparities in the rates of stops . . . of African Americans.” Id. at 3, 5–6, 48–74.
Yet these discriminatory practices “uncover[ed] evidence of criminal activity” at an
“extremely low rate.” Id. at 28; see, e.g., id. at 50 (“One African-American man in his
mid-fifties was stopped 30 times in less than four years” based only on a suspicion of
“‘loitering’ or ‘trespassing’” and “none of the 30 stops resulted in a citation or criminal
charge”). Although DOJ’s investigation resulted in a federal consent decree with the BPD,
discriminatory policing continues to affect city residents. 13

13

See, e.g., Morgan State Univ. Inst. for Urban Research, The Community’s
Experiences and Perceptions of the Baltimore City Police Department Survey Report 8–21
(2020), https://perma.cc/WX8D-GXX8 (finding, in survey conducted as part of federal
consent decree monitoring, that a majority of respondents had personally witnessed BPD
officers engaged in racial profiling, among other findings of community dissatisfaction)
(saved as ECF opinion attachment); Justin Fenton, Corrupt Squad Scoured Baltimore
Streets in Pursuit of Black Men to Search, Arrest—and Steal From, Baltimore Sun (June
12, 2019), https://perma.cc/2W8Q-44VN (reporting on the now-infamous “Gun Trace
Task Force” scandal involving a unit of BPD officers who routinely planted false evidence
and stole from vulnerable community members, continuing even through DOJ’s
investigation and the imposition of the federal consent decree) (saved as ECF opinion
attachment); see also Brief of Amici Curiae Victims of the Baltimore Police Department
in Support of Appellee at 6–26, Baltimore City Police Department v. Potts, Misc. No. 6
(Md. Dec. 16, 2019), available at https://perma.cc/ZLA2-H6ZY (collecting evidence to
show that the conduct underlying the Gun Trace Task Force scandal more broadly pervades
BPD’s policing and continues through the present-day) (saved as ECF opinion attachment).
47

 This recent history provides important context to the majority’s conclusions that
BPD has “consulted with the people of Baltimore,” such that the AIR program is ultimately
“a community initiative.” See Maj. Op. at 22. It counsels that the consequences of the AIR
program’s unconstitutional nature are likely to be felt in the same overpoliced
neighborhoods where BPD’s discriminatory practices have already left generational
legacies. 14
14

In particular, several of the historically Black neighborhoods in West Baltimore
have been shaped by decades of segregationist and exclusionary policy and policing. See
DOJ Report at 12–15, 70–73. Baltimore was the first city in America to “establish[] blockby-block segregation,” enforced over the decades by tactics including “restrictive
covenants, aggressive redlining, a contract system for housing loans, and racially targeted
subprime loans.” Id. at 12. As a result, “Baltimore remains among the most segregated
cities in the country.” Id.
For example, the West Baltimore neighborhood of Sandtown-Winchester—now
marked in history as the home of Freddie Gray—has been at least 90% Black “for the past
five census periods,” and is currently more than 95% Black. Id. at 13, 16–18 (citing No
Boundaries Coalition, Over-Policed, Yet Underserved (“Over-Policed Report”), March
2016, https://perma.cc/KD7V-PZRG) (saved as ECF opinion attachment).
The
neighborhood has “a rich history of arts, culture, and civil rights activism”; “[p]erformers
such as Billie Holliday and Diana Ross performed at venues along the historic
Pennsylvania Avenue, often called ‘Harlem of the South,’” and “‘The Avenue’ was known
as the heart of Baltimore’s Entertainment scene during the time of prohibition through the
civil rights movement.” Over-Policed Report 6.
But through the mid-twentieth century, as Baltimore’s industrial sector declined and
middle-class “flight” to the suburbs proliferated, Sandtown suffered from “economic
depression, housing abandonment, [and] increased crime.” Id. Today, one-in-three of the
neighborhood’s houses are vacant or abandoned, more than one-fifth of residents are
unemployed, and more residents are imprisoned than those of any other Baltimore
neighborhood. Outlaw, supra, at 38. Life expectancy in Sandtown is 69.7 years, compared
to 84.4 years in the majority-white neighborhood Roland Park. Id. The Sandtown infant
mortality rate is 9.7 per 1,000 births, compared to Roland Park’s 3.4, and Sandtown has
been the neighborhood most heavily impacted by Baltimore’s long-standing lead poisoning
epidemic. Id.; DOJ Report at 14.
(Continued)
48

 Yet the majority reasons that—“[a]s cherished as constitutional rights are”—“the
Bill of Rights was not the only source of rights worthy of protection.” Maj. Op. at 23.
“When hundreds of Baltimore residents are killed on their streets each year,” or “[w]hen
criminals can rob Baltimoreans at gunpoint with apparent impunity,” their “rights to life,”
“rights to liberty,” and “rights to property” are not protected. Maj. Op. at 23. Therefore,
“Baltimore and its citizens have a strong interest in the AIR program not being strangled
in its infancy.” Maj. Op. at 22.
Under this line of reasoning, which rights—and whose rights—are regarded as
expendable in the name of protecting life, liberty, and property? Why must the people of
Baltimore, that the majority regards as so thoroughly victimized by crime and violence,
“decide between their constitutional rights against unwarranted searches and seizures or
forgo governmental protection that is readily afforded to other communities[?]” Curry,
965 F.3d at 333 (Gregory, J., concurring).

Yet while Sandtown remained under-resourced, it became over-policed. See
Over-Policed Report 10–19; see also DOJ Report at 51 (showing, on a map, how the
frequency of police stops correlates with the Blackness of a given Baltimore neighborhood,
with Sandtown among the highest density of stops). Under BPD’s “zero tolerance”
approach to policing, implemented during its “War on Drugs,” the police prioritized
“broken windows” strategies, “stop and frisk” searches, and brutal physical confrontations.
Over-Policed Report 11–12. When Sandtown residents have requested a more supportive
police presence in the neighborhood they have been rebuffed, while predominantly-White
neighborhoods received those same services. See id.; DOJ Report 4–5, 65–66, 156. Thus,
residents report “that the legacy of racism in Baltimore is a defining feature of community
life and is experienced through concentrated poverty, disinvestment, discrimination, and
police profiling and abuse in Sandtown.” Over-Policed Report 19. In neighborhoods like
Sandtown, trauma caused by policing has made a “transgenerational impact” with
“long-term consequences . . . [that] cannot be understated or ignored.” See Over-Policed
Report 19. This is the living and breathing history that should inform our understanding
of the consequences of police practices that trample the Fourth Amendment’s protections.
49

 No crime rate can justify warrantless aerial surveillance of an entire city, wholly
unchecked by the judiciary. This case presents another circumstance in which “we must
remind law enforcement that the Fourth Amendment protects against unreasonable
searches and seizures” to prevent “the slow systematic erosion of Fourth Amendment
protections for a certain demographic.” United States v. Black, 707 F.3d 531, 534, 542
(4th Cir. 2013). Any innovation by the police must occur within the confines of the
Constitution. C.f. Curry, 965 F.3d at 345 (Thacker, J., concurring) (“Technology cannot
override human flaws.”).
I cast no aspersions on the majority’s sincere concerns regarding the problem of
crime in the city of Baltimore. “Certainly, the law-abiding residents of Baltimore want and
welcome policing. They want safe streets for their families, to feel secure in their homes,
and to go about their lives without being robbed, assaulted, or killed.” Outlaw, supra, at
40. “However, the manner in which policing occurs is just as important as the level of
policing.” Id. Here, the manner selected by BPD violates the Constitution.
However, the majority posits that halting the AIR program—in the face of
“hundreds of people being murdered every year”—would acquiesce to “some new normal
of indifference.” Maj. Op. at 23. Granting plaintiffs an injunction would leave the City
“with lessened hope and not knowing where to turn.” Maj. Op. at 22.
But Baltimore is not a hopeless place. One need only look to the plaintiffs
themselves to find proof of that. For example, Plaintiff Erricka Bridgeford is a longtime
activist in Baltimore, most prominently as an organizer behind Ceasefire Baltimore 365,
an organization devoted to ending gun violence in the City. Ms. Bridgeford has organized
50

 high-profile “ceasefire weekends” and personally visits every murder site within the city
shortly after violence takes place. 15
When hearing about Baltimore’s crime statistics, Ms. Bridgeford has responded:
“But the thing about Baltimore is that it has never been the one to just be defeated. So the
story about the Baltimore ceasefire is that Baltimore looked the murder rate in the eye and
said, ‘What you’re not going to do is snatch our greatness.’” Erricka Bridgeford, How
Baltimore Called a Ceasefire, TED (Oct. 2017), https://perma.cc/Q97D-D89G (saved as
ECF opinion attachment). In fact, Ms. Bridgeford was moved to begin organizing ceasefire
weekends specifically in response to hearing that common refrain about the city’s high
murder rate. “[B]ecause my city looks broken and is always yearning to show up in its
wholeness,” she knew that “people around this city wanted to do something great together.”
Id. And through her work, Ms. Bridgeford “ha[s] seen the power of collective consciousness.
. . . [P]eople around this world mischaracterized [Baltimore] and misunderstood it. What you
failed to realize is Baltimore has the power to rise up, and that is what we continue to do.” Id.
Ms. Bridgeford’s leadership shows that—regardless of this Court’s dicta—the people of
Baltimore do, in fact, know where to turn. They turn to each other.
This Court should not invoke the tragedies imparted by gun violence in Baltimore
to justify its grant of sweeping surveillance powers to the BPD. This Court should not

15

One study—cited in Ms. Bridgeford’s declaration to the district court—found a
52% reduction in gun violence in Baltimore City during the ceasefire weekends, with no
evidence of a “postponement effect” in the days immediately following. Peter Phaylen et
al., Baltimore Ceasefire 365: Estimated Impact of a Recurring Community-Led Ceasefire
on Gun Violence, 110 Am. J. Pub. Health 554, 554 (2020), https://perma.cc/MZ79-82EC
(saved as ECF opinion attachment).
51

 “[p]ermit[] unconstitutional governmental intrusions into these communities in the name
of protecting them.” Curry, 965 F.3d at 332 (Gregory, J., concurring). “If merely
preventing crime was enough to pass constitutional muster, the authority of the Fourth
Amendment would become moot.” Id. at 334.
The dragnet aerial surveillance program implemented by the BPD violates the
Fourth Amendment, and the plaintiffs are entitled to a preliminary injunction to halt its
operation. Meanwhile, in the face of strife and struggle, the people of Baltimore will
continue to turn to their resiliency, their beauty, and their hope.

52